[ { "caveats": [ "Rented single-GPU class; model and engine pins were not retained. Historical TTFT p95 withdrawn because raw samples cannot reconstruct corrected percentiles.", "Publisher revision was not recorded for public reproduction.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "vLLM", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "5cae73717fc898b2755a8d10", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 168 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 168 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 883.3 }, { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "median", "unit": "tok/s", "value": 30.7 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen2.5-7B", "publisher_revision": null, "variant": "unrecorded variant", "weights_dtype": "unrecorded", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "168/168" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-07-16", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 8192, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "62a5a7d509a66a80aef56e64394d1250", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA A10, 24 GB", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Rented single-GPU class; model and engine pins were not retained. Historical TTFT p95 withdrawn because raw samples cannot reconstruct corrected percentiles.", "Publisher revision was not recorded for public reproduction.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "vLLM", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "6ccd93b0e0d9a814c44d74b6", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 168 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 168 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 1261.9 }, { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "median", "unit": "tok/s", "value": 43.8 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen2.5-72B-Instruct-AWQ", "publisher_revision": null, "variant": "AWQ variant", "weights_dtype": "AWQ", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "168/168" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-07-17", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 8192, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "8a74ae0259361994e9271094679763cd", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA H100, 80 GB", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "No successful inference result was produced; this failed attempt remains official evidence.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "vLLM", "recipe": "plain", "version": "0.26.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "26af7cc2bed86f8fcf33bd65", "measured_at": "2026-10-04T10:43:24.538266+00:00", "metrics": [], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [ "Image identity could not be verified." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-04", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": { "context_tokens": 8192, "speculative_decoding": "plain" }, "setup_complete": false, "setup_id": "2c2439dfecffd5beba82d0ac4211451f", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "No successful inference result was produced; this failed attempt remains official evidence.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "vLLM", "recipe": "plain", "version": "0.26.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "290d4246df1c6aa7fa2b7f63", "measured_at": "2026-10-04T19:07:35.739229+00:00", "metrics": [], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [ "Runtime did not become healthy." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-04", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": { "context_tokens": 8192, "speculative_decoding": "plain" }, "setup_complete": false, "setup_id": "2c2439dfecffd5beba82d0ac4211451f", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Recorded seed, not a fresh rerun; service targets and broad task quality unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "3671f7ed4a19fa2721ae1f21", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "single", "unit": "tok/s", "value": 14.3 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-04", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 8192, "speculative_decoding": "plain" }, "setup_complete": false, "setup_id": "0bd1e7b070b83f21453a42d457c59012", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Short repeated prompts; prefix-slot reuse may improve TTFT. Task correctness and long-context quality unassessed; whole-host memory includes baseline.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp", "recipe": "with multi-token prediction", "version": "native CUDA 13, sm_121a" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "08b6b6698d9424baa272c369", "measured_at": "2026-10-05T18:20:22.433708+00:00", "metrics": [ { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 26.98740543967368 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_ttft_mean_s", "scope": "per_person", "source_field": "code_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.29756888799602166 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 53.585154787958096 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 33.642655882501664 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.2478214856625224 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 27.523632049560547 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "GSQ-RCO IQ3_S GGUF variant", "weights_dtype": "GSQ-RCO IQ3_S", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 2, "context_tokens": 65536, "max_output_tokens": 512 }, "setup_complete": false, "setup_id": "7b2be9be996767ee320ef3e0abcb3035", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Failure retained. Earlier images and request settings differ from the final drive; no package promotion.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "26ea705e3f3b6841efe9974c", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 30 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[0].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[0].passed", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[0].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[1].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[1].passed", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[1].total", "statistic": "single", "unit": "count", "value": 16 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "structured-tasks", "observed": "30/32" } ], "reasons": [ "context-specific-exception returned an extra citation at concurrency 1 and fenced JSON at concurrency 2" ], "status": "partial" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 30, "skipped": null, "suite": "structured-tasks", "total": 32, "version": null } ], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "bcdf385cd1ff43923b9a7bfaf9fb1df0", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Failure retained. Earlier images and request settings differ from the final drive; no package promotion.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "2c5a80f024624c17ca9ac701", "measured_at": null, "metrics": [], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [ "Pinned ServerArgs is immutable after resolution; credentials now enter the in-process argument parser before resolution" ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "bcdf385cd1ff43923b9a7bfaf9fb1df0", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab measurement; broader quality and customer acceptance remain unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "33a18763535def254c3e5a1f", "measured_at": "2026-10-05T03:27:00.672910+00:00", "metrics": [ { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 16 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "structured-tasks", "observed": "16/16" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 16, "skipped": null, "suite": "structured-tasks", "total": 16, "version": null } ], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": { "context_tokens": 8192, "speculative_decoding": "plain" }, "setup_complete": false, "setup_id": "0bd1e7b070b83f21453a42d457c59012", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Short repeated code/chat prompts; output hit the token ceiling; task correctness and full context quality unassessed.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "vLLM", "recipe": "speculative decoding", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "452eb52c2cbcb1e0e445543b", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 27.763940270787135 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 43.445051336045445 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 120.29696655273438 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "speculative decoding" }, "setup_complete": false, "setup_id": "daff90b35f53910d756e362003f54ed8", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Bounded lab package proof only; configured 262144 context is not long-context quality. Memory is sampled whole-host unavailable memory during speed, not allocator residency or the entire drive peak. Candidate: one concurrency-2 answer returned fenced JSON and failed the unchanged strict contract. Uncalibrated FP8 KV scales remain a quality caveat.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "4d30da78f0b56f97709f43d6", "measured_at": "2026-10-05T22:21:12.790316+00:00", "metrics": [ { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": 3, "name": "chat_decode_tps", "samples": [ 34.347989734983216, 34.369737142787336, 34.28250445000213 ], "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "stddev": 0.045407075413807445, "unit": "tok/s", "value": 34.333410442590896 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": 3, "name": "code_ttft_mean_s", "samples": [ 0.19521947705652565, 0.19531845301389694, 0.19371042301645502 ], "scope": "per_person", "source_field": "code_ttft_mean_s", "statistic": "mean", "stddev": 0.0009011844806201899, "unit": "s", "value": 0.19474945102895921 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[0].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[0].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[0].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[1].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[1].passed", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[1].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 56.92073838049442 }, { "basis": null, "n": 3, "name": "code_decode_tps", "samples": [ 57.62747328933936, 57.454320212510794, 57.50305099811821 ], "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "stddev": 0.08929126873500154, "unit": "tok/s", "value": 57.528281499989454 }, { "basis": null, "n": 3, "name": "chat_ttft_mean_s", "samples": [ 0.1922413359861821, 0.19155225198483095, 0.1926738429465331 ], "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "stddev": 0.0005656655839277865, "unit": "s", "value": 0.19215581030584872 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 113.64101028442383 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" }, { "met": false, "name": "structured-tasks", "observed": "15/16" } ], "reasons": [ "15/16 tasks passed." ], "status": "partial" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "structured-tasks", "total": 16, "version": null } ], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": "446eae1b7d949003bd1379b1d1614c10b4f484a093cea9a3dbab7db56672fe5b", "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": { "concurrency": 1, "context_tokens": 262144, "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "speculative_decoding": "speculative decoding" }, "setup_complete": false, "setup_id": "bcdf385cd1ff43923b9a7bfaf9fb1df0", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Short repeated code/chat prompts; output hit the token ceiling; task correctness and full context quality unassessed.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "vLLM", "recipe": "speculative decoding", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "60cee206868b13f9f5272ee1", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 27.698653775158036 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 45.22025818959093 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 92.43753051757812 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant with BF16 head", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "speculative decoding" }, "setup_complete": false, "setup_id": "be11087c29d0d420b8308f677c8ff795", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Short repeated code/chat prompts; output hit the token ceiling; task correctness and full context quality unassessed. Missing FP8 KV scales defaulted to 1.0; accuracy unassessed.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "62294ed0cfe43fd8752403de", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 26.43902638526529 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 40.56478674951746 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 112.65891647338867 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant with BF16 head", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 262144, "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "speculative_decoding": "speculative decoding" }, "setup_complete": false, "setup_id": "8e9b94c77df1135117319c5aa6b48933", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Recorded seed, not a fresh rerun; service targets and broad task quality unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "6d39ad79aafbfdac807c2434", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "single", "unit": "tok/s", "value": 23.3 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 8192, "speculative_decoding": "with multi-token prediction" }, "setup_complete": false, "setup_id": "0441728d4c7a1e06d33c6bfe4dffb271", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Short repeated prompts; prefix-slot reuse may improve TTFT. Task correctness and long-context quality unassessed; whole-host memory includes baseline.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp", "recipe": "with multi-token prediction", "version": "native CUDA 13, sm_121a" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "71cdb415e8151af4d03692cb", "measured_at": "2026-10-05T17:54:31.832107+00:00", "metrics": [ { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 31.589469936120086 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_ttft_mean_s", "scope": "per_person", "source_field": "code_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.13722664832797213 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 38.03795761378485 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.13756801267542565 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 21.732799530029297 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "GSQ-RCO IQ3_S GGUF variant", "weights_dtype": "GSQ-RCO IQ3_S", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 65536, "max_output_tokens": 512 }, "setup_complete": false, "setup_id": "7b2be9be996767ee320ef3e0abcb3035", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Short repeated code/chat prompts; output hit the token ceiling; task correctness and full context quality unassessed. Missing FP8 KV scales defaulted to 1.0; accuracy unassessed.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "75a65a840c89fbde7ce3c663", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 19.732668519416297 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 24.14055995668283 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 110.51219940185547 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant with BF16 head", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 262144, "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "speculative_decoding": "speculative decoding" }, "setup_complete": false, "setup_id": "4f1055474916fa6c413b6b9557094b64", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Recorded seed, not a fresh rerun; service targets and broad task quality unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "7abd98af66d5ce35dfa17e71", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 16 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "16/16" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 8192, "speculative_decoding": "plain" }, "setup_complete": false, "setup_id": "0bd1e7b070b83f21453a42d457c59012", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Earlier attempt; comparison or runtime-boundary checks stopped the sweep. No winner promoted. Every attempted and unrun candidate is retained.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "810e107835d74649bd4219a4", "measured_at": "2026-10-05T15:22:33.276063+00:00", "metrics": [ { "basis": null, "n": null, "name": "tuning_candidates", "scope": "machine_total", "source_field": "tuning_candidates", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "rejected_candidates", "scope": "machine_total", "source_field": "rejected_candidates", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[0].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "trials[0].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "trials[0].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "trials[0].aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 41.5929854435505 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[1].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[2].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[3].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[4].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[5].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[6].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[7].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[8].concurrency", "statistic": "single", "unit": "count", "value": 2 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [ "Earlier attempt; comparison or runtime-boundary checks stopped the sweep." ], "status": "partial" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "auto-tune", "version": null }, "settings": { "context_tokens": 262144, "speculative_decoding": "speculative decoding" }, "setup_complete": false, "setup_id": "5dc339b5547410253a1a8f37c086747b", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Two identical-seed images are a bounded smoke check, not general image quality. CUDA allocator and whole-host memory measure different scopes.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Keep image engine on NVIDIA NGC PyTorch 25.11 + Diffusers 0.40", "recipe": "plain", "version": "Diffusers 0.40.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "8426a04db8f9fafa5bad5c1c", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "peak_allocated_gib", "scope": "machine_total", "source_field": "peak_allocated_gib", "statistic": "max", "unit": "GiB", "value": 54.16329288482666 }, { "basis": null, "n": null, "name": "seconds_per_image", "scope": "per_person", "source_field": "seconds_per_image", "statistic": "mean", "unit": "s/image", "value": 159.01967584650265 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "peak_reserved_gib", "scope": "machine_total", "source_field": "peak_reserved_gib", "statistic": "max", "unit": "GiB", "value": 54.673828125 }, { "basis": null, "n": null, "name": "images", "scope": "machine_total", "source_field": "images", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 62.49386215209961 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen-Image-2512", "publisher_revision": "25468b98e3276ca6700de15c6628e51b7de54a26", "variant": "BF16 variant", "weights_dtype": "BF16", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "structured-tasks", "observed": "2/2" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 2, "skipped": null, "suite": "structured-tasks", "total": 2, "version": null } ], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "media-image", "version": null }, "settings": { "dtype": "bfloat16", "height": 1024, "speculative_decoding": "plain", "width": 1024 }, "setup_complete": false, "setup_id": "1813e14b6165264ca54b389ad3db1c6f", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Short repeated code/chat prompts; output hit the token ceiling; task correctness and full context quality unassessed.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "vLLM", "recipe": "speculative decoding", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "85224459e5f303f76fb77b2c", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 25.415653737014367 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 42.74614542052032 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 92.13587188720703 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant with BF16 head", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 2, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "speculative decoding" }, "setup_complete": false, "setup_id": "be11087c29d0d420b8308f677c8ff795", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Short repeated prompts; prefix-slot reuse may improve TTFT. Task correctness and long-context quality unassessed; whole-host memory includes baseline.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp", "recipe": "plain", "version": "native CUDA 13, sm_121a" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "878d422a9fda08591057e990", "measured_at": "2026-10-05T17:59:29.228107+00:00", "metrics": [ { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 16.04749708252017 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_ttft_mean_s", "scope": "per_person", "source_field": "code_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.13250124733895063 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 15.958922167679104 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.13248725267476402 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 20.367591857910156 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "GSQ-RCO IQ3_S GGUF variant", "weights_dtype": "GSQ-RCO IQ3_S", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 65536, "max_output_tokens": 512 }, "setup_complete": false, "setup_id": "0ec4454b3b077e7c71832164cf1f80eb", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Short repeated code/chat prompts; output hit the token ceiling; task correctness and full context quality unassessed.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "8e72371cd9b0130995d39315", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 29.155003923596308 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 52.674847621473496 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 91.91862869262695 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 2, "context_tokens": 262144, "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "speculative_decoding": "speculative decoding" }, "setup_complete": false, "setup_id": "9275dcbc479d02a0b4f4be004f7bba2a", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Two identical-seed images are a bounded smoke check, not general image quality. CUDA allocator and whole-host memory measure different scopes.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Keep image engine on NVIDIA NGC PyTorch 25.11 + Diffusers 0.40", "recipe": "plain", "version": "Diffusers 0.40.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "adda7e6d920481943774f2fd", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "peak_allocated_gib", "scope": "machine_total", "source_field": "peak_allocated_gib", "statistic": "max", "unit": "GiB", "value": 21.669125080108643 }, { "basis": null, "n": null, "name": "seconds_per_image", "scope": "per_person", "source_field": "seconds_per_image", "statistic": "mean", "unit": "s/image", "value": 12.950703782495111 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "peak_reserved_gib", "scope": "machine_total", "source_field": "peak_reserved_gib", "statistic": "max", "unit": "GiB", "value": 24.4609375 }, { "basis": null, "n": null, "name": "images", "scope": "machine_total", "source_field": "images", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 33.77626037597656 } ], "model": { "artifact_sha256": [], "publisher_repo": "Tongyi-MAI/Z-Image-Turbo", "publisher_revision": "f332072aa78be7aecdf3ee76d5c247082da564a6", "variant": "BF16 variant", "weights_dtype": "BF16", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "structured-tasks", "observed": "2/2" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 2, "skipped": null, "suite": "structured-tasks", "total": 2, "version": null } ], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "media-image", "version": null }, "settings": { "dtype": "bfloat16", "height": 1024, "speculative_decoding": "plain", "width": 1024 }, "setup_complete": false, "setup_id": "a9f70cba4e423cd43c97a7eef6de854e", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Failure retained. Earlier images and request settings differ from the final drive; no package promotion.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "b749797cadad45718f30c3c9", "measured_at": null, "metrics": [], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [ "CUDA allocation failed during context initialization before weight loading. No other engine was running. A later retry after per-file cache advice started successfully; this does not establish the sole cause." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "bcdf385cd1ff43923b9a7bfaf9fb1df0", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Configured 8K context only; no declared customer service targets. Earlier failed smoke remains in the run history.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "c8b7c076e64b38b83c820637", "measured_at": "2026-10-05T05:54:15.517300+00:00", "metrics": [ { "basis": null, "n": null, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 0.814 }, { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "ttft_p95_s", "scope": "per_person", "source_field": "ttft_p95_s", "statistic": "p95", "unit": "s", "value": 0.815 }, { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.6 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "structured-tasks", "observed": "16/16" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 16, "skipped": null, "suite": "structured-tasks", "total": 16, "version": null } ], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": { "concurrency": 1, "context_tokens": 8192, "kv_dtype": "bf16", "speculative_decoding": "plain" }, "setup_complete": false, "setup_id": "0bd1e7b070b83f21453a42d457c59012", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Short repeated code/chat prompts; output hit the token ceiling; task correctness and full context quality unassessed. Missing FP8 KV scales defaulted to 1.0; accuracy unassessed.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "d43d3d2921d854178c4193f4", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 21.238341518098036 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 31.991125012719905 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 115.10027694702148 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant with BF16 head", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 262144, "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "speculative_decoding": "speculative decoding" }, "setup_complete": false, "setup_id": "ae7164bf7eef4373dec8f83258848782", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "The selected candidate passed its bounded checks; rejected and unfinished candidates remain in the table. Configured context is not long-context quality proof.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "d732a4ff7cc1dd9ae54bd9e0", "measured_at": "2026-10-05T16:30:42.252659+00:00", "metrics": [ { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "tuning_candidates", "scope": "machine_total", "source_field": "tuning_candidates", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "rejected_candidates", "scope": "machine_total", "source_field": "rejected_candidates", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 59.1631463518914 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[0].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "trials[0].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "trials[0].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "trials[0].aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 41.21903238019381 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[1].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "trials[1].passed", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "trials[1].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "trials[1].aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 37.8443178159562 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[2].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "trials[2].passed", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "trials[2].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "trials[2].aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 60.839773492013876 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[3].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "trials[3].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "trials[3].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "trials[3].aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 37.939468739670176 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[4].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "trials[4].passed", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "trials[4].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "trials[4].aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 61.69623714961223 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[5].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "trials[5].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "trials[5].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "trials[5].aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 41.32935955646878 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[6].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "trials[6].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "trials[6].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "trials[6].aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 57.71034425957966 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[7].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "trials[7].passed", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "trials[7].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "trials[7].aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 38.401894829174765 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[8].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "trials[8].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "trials[8].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "trials[8].aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 59.1631463518914 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "structured-tasks", "observed": "16/16" } ], "reasons": [ "The selected candidate passed its bounded checks; rejected and unfinished candidates remain in the table." ], "status": "partial" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 16, "skipped": null, "suite": "structured-tasks", "total": 16, "version": null } ], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "auto-tune", "version": null }, "settings": { "concurrency": 2, "context_tokens": 262144, "kv_dtype": "fp8_e4m3", "speculative_decoding": "speculative decoding" }, "setup_complete": false, "setup_id": "5dc339b5547410253a1a8f37c086747b", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Short repeated code/chat prompts; output hit the token ceiling; task correctness and full context quality unassessed.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "d806de800b927c6b0ef8d803", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 29.692044188605895 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 57.95064400812194 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 91.60038757324219 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 262144, "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "speculative_decoding": "speculative decoding" }, "setup_complete": false, "setup_id": "9275dcbc479d02a0b4f4be004f7bba2a", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "15/16 at both concurrency levels; measured but unqualified. Short prompts and reused caches do not establish long-context quality.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp", "recipe": "with multi-token prediction", "version": "native CUDA 13, sm_121a" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "d92bd079da7d9fc5722fc394", "measured_at": "2026-10-05T18:22:43.041657+00:00", "metrics": [ { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 26.98740543967368 }, { "basis": null, "n": null, "name": "code_ttft_mean_s", "scope": "per_person", "source_field": "code_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.29756888799602166 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[0].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[0].passed", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[0].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "ttft_p95_s", "scope": "per_person", "source_field": "workload_levels[0].ttft_p95_s", "statistic": "p95", "unit": "s", "value": 0.5690766500192694 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[1].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[1].passed", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[1].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "ttft_p95_s", "scope": "per_person", "source_field": "workload_levels[1].ttft_p95_s", "statistic": "p95", "unit": "s", "value": 0.794703905005008 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 53.585154787958096 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 33.642655882501664 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.2478214856625224 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 27.523632049560547 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "GSQ-RCO IQ3_S GGUF variant", "weights_dtype": "GSQ-RCO IQ3_S", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": false, "name": "structured-tasks", "observed": "15/16" } ], "reasons": [ "One task returned fenced text instead of one strict JSON value at each concurrency level.", "Native runtime identity does not satisfy the workload runner's OCI identity gate." ], "status": "partial" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "structured-tasks", "total": 16, "version": null } ], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": { "concurrency": 2, "context_tokens": 65536, "max_output_tokens": 512 }, "setup_complete": false, "setup_id": "7b2be9be996767ee320ef3e0abcb3035", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Tooling failure; the corrected startup and subsequent strict workload score are retained separately.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp", "recipe": "with multi-token prediction", "version": "native CUDA 13, sm_121a" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "e07e13bacf97b8d0559da6be", "measured_at": null, "metrics": [], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "GSQ-RCO IQ3_S GGUF variant", "weights_dtype": "GSQ-RCO IQ3_S", "weights_format": "GGUF" }, "outcome": { "gates": [], "reasons": [ "The initial workload attempt was refused before tasks because authentication was missing." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "7b2be9be996767ee320ef3e0abcb3035", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Tooling failure; the corrected startup and subsequent strict workload score are retained separately.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp", "recipe": "with multi-token prediction", "version": "native CUDA 13, sm_121a" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "ee7ff82a323045f554e5467f", "measured_at": null, "metrics": [], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "GSQ-RCO IQ3_S GGUF variant", "weights_dtype": "GSQ-RCO IQ3_S", "weights_format": "GGUF" }, "outcome": { "gates": [], "reasons": [ "The initial native MTP launch rejected an unsupported command option." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "7b2be9be996767ee320ef3e0abcb3035", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Configured 8K context only; no declared customer service targets. Workload passed 16/16, but the optional smoke failed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "f4c10f9c358d36fba0d11619", "measured_at": "2026-10-05T04:54:24.359517+00:00", "metrics": [ { "basis": null, "n": null, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 8.599987004010472 }, { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "ttft_p95_s", "scope": "per_person", "source_field": "ttft_p95_s", "statistic": "p95", "unit": "s", "value": 11.441950695007108 }, { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.672248485179336 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "structured-tasks", "observed": "16/16" } ], "reasons": [ "Optional smoke exhausted its output budget before a final answer." ], "status": "partial" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 16, "skipped": null, "suite": "structured-tasks", "total": 16, "version": null } ], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": { "concurrency": 1, "context_tokens": 8192, "kv_dtype": "bf16", "speculative_decoding": "plain" }, "setup_complete": false, "setup_id": "0bd1e7b070b83f21453a42d457c59012", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Earlier attempt; comparison or runtime-boundary checks stopped the sweep. No winner promoted. Every attempted and unrun candidate is retained.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "f8a9c1e519df761b52177b08", "measured_at": "2026-10-05T15:43:15.587639+00:00", "metrics": [ { "basis": null, "n": null, "name": "tuning_candidates", "scope": "machine_total", "source_field": "tuning_candidates", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "rejected_candidates", "scope": "machine_total", "source_field": "rejected_candidates", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[0].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "trials[0].passed", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "trials[0].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "trials[0].aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 38.57618374487346 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[1].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "trials[1].passed", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "trials[1].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "trials[1].aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 37.99076401854043 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[2].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "trials[2].passed", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "trials[2].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "trials[2].aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 60.15292034760066 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[3].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[4].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[5].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[6].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[7].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "trials[8].concurrency", "statistic": "single", "unit": "count", "value": 2 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [ "Earlier attempt; comparison or runtime-boundary checks stopped the sweep." ], "status": "partial" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-05", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "auto-tune", "version": null }, "settings": { "context_tokens": 262144, "speculative_decoding": "speculative decoding" }, "setup_complete": false, "setup_id": "5dc339b5547410253a1a8f37c086747b", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "041ce71d9e1a4100abecb8d7", "measured_at": "2026-10-06T07:01:56.682938+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 730.6092431609868 }, { "basis": null, "n": 2, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 546.450626095495 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": 2, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 724.1910616810201 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 368.71019050996983 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 730.2259663100122 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.2434195293559571 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 678.0691351497641 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 89 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 724.1910616810201 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 730.6011252800236 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 15.912478624370767 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 345.22795603091936 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.5250204397813543 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.26279437578603637 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 116.31097412109375 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "2/2" }, { "met": true, "name": "hidden-facts", "observed": "6/6" }, { "met": false, "name": "first-token-300-s-reference", "observed": 724.1910616810201 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 6, "skipped": null, "suite": "hidden-facts", "total": 6, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.2434195293559571, "effective_prefill_tps": 678.0691351497641, "needle_correct": 3, "needle_total": 3, "output_tokens": 89, "prompt_tokens": 250011, "total_s": 730.2259663100122, "ttft_s": 368.71019050996983, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 15.912478624370767, "effective_prefill_tps": 345.22795603091936, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 730.6011252800236, "ttft_s": 724.1910616810201, "user_number": 1, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 2, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "10597aa97c3a6a4e114a3fa137d2bc6c", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011.0, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "055e55e7006f46ae9559475d", "measured_at": "2026-10-06T07:46:35.891108+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 1456.9594599839766 }, { "basis": null, "n": 4, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 911.6945878329861 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": 4, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 1446.3931534200092 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 1446.3931534200092 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 1455.5834148149588 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.0987049896157 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 172.8513436397613 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 1091.54475121398 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 1456.4609519059886 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.28773729366052614 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 229.0432890836093 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 106 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 731.8444244519924 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 1456.7376081240363 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.1476079543995394 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 341.6176876488594 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 368.89545817900216 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 1456.951508337981 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.09925959235668036 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 677.7285934452603 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.387847666210278 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.29238974158188663 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 116.28083801269531 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "4/4" }, { "met": true, "name": "hidden-facts", "observed": "12/12" }, { "met": false, "name": "first-token-300-s-reference", "observed": 1446.3931534200092 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 12, "skipped": null, "suite": "hidden-facts", "total": 12, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 11.0987049896157, "effective_prefill_tps": 172.8513436397613, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1455.5834148149588, "ttft_s": 1446.3931534200092, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.28773729366052614, "effective_prefill_tps": 229.0432890836093, "needle_correct": 3, "needle_total": 3, "output_tokens": 106, "prompt_tokens": 250011, "total_s": 1456.4609519059886, "ttft_s": 1091.54475121398, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.1476079543995394, "effective_prefill_tps": 341.6176876488594, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "total_s": 1456.7376081240363, "ttft_s": 731.8444244519924, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.09925959235668036, "effective_prefill_tps": 677.7285934452603, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 1456.951508337981, "ttft_s": 368.89545817900216, "user_number": 3, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 4, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "10597aa97c3a6a4e114a3fa137d2bc6c", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011.0, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 4, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "063d452cce7348478144544b", "measured_at": "2026-10-06T03:06:31.741169+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 1395.1257285339525 }, { "basis": null, "n": 3, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 932.0302819130011 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": 3, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 1379.8438422949985 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 471.9909305470064 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 1386.734988627024 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.1115066002331742 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 529.6945000834098 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 932.0302819130011 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 1391.490379876981 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.22199969149006807 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 268.24343033881905 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 1379.8438422949985 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 1395.084906300006 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.692446141981135 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 181.18789412008684 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.3314937652694489 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.2214854143108006 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 103.3937759399414 } ], "model": { "artifact_sha256": [ "1d8268aa85ace093a561e3e7b63b9d390dac1cd55a90cd55b5ec509c3c9da9fe", "f078f51aed0c1dba613c1759487004ee06aba1a290606b6be125f59ab710959e" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "3/3" }, { "met": true, "name": "hidden-facts", "observed": "9/9" }, { "met": false, "name": "first-token-300-s-reference", "observed": 1379.8438422949985 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 9, "skipped": null, "suite": "hidden-facts", "total": 9, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.1115066002331742, "effective_prefill_tps": 529.6945000834098, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1386.734988627024, "ttft_s": 471.9909305470064, "user_number": 2, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.22199969149006807, "effective_prefill_tps": 268.24343033881905, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1391.490379876981, "ttft_s": 932.0302819130011, "user_number": 0, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 6.692446141981135, "effective_prefill_tps": 181.18789412008684, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1395.084906300006, "ttft_s": 1379.8438422949985, "user_number": 1, "valid_json": null } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 3, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "2a73804cb33351a1318b0917ba061450", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 3, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "0fef1bb055d946f69b0101c6", "measured_at": "2026-10-06T04:17:32.449747+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 2323.0951668579946 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 1395.2731267259805 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 2307.5008943669964 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 473.0587294619763 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 1915.9567166739726 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.07069106818638438 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 528.4988616198773 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 933.6912666810094 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 2308.7227476239786 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.07418011981082094 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 267.7662402141916 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 1395.2731267259805 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 2314.8470665949862 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.1120083938162411 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 179.18427239163762 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 104 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 1858.0508087389753 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 2319.455895067018 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.22106388295746185 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 134.55552389855143 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 2307.5008943669964 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 2323.056761979009 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.557011318432206 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 108.34708693301899 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.2762164862222991 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.22211746094667928 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 103.49026107788086 } ], "model": { "artifact_sha256": [ "1d8268aa85ace093a561e3e7b63b9d390dac1cd55a90cd55b5ec509c3c9da9fe", "f078f51aed0c1dba613c1759487004ee06aba1a290606b6be125f59ab710959e" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 2307.5008943669964 } ], "reasons": [ "Observed decode overlap was 4 of 5 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.07069106818638438, "effective_prefill_tps": 528.4988616198773, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1915.9567166739726, "ttft_s": 473.0587294619763, "user_number": 0, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.07418011981082094, "effective_prefill_tps": 267.7662402141916, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 2308.7227476239786, "ttft_s": 933.6912666810094, "user_number": 3, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.1120083938162411, "effective_prefill_tps": 179.18427239163762, "needle_correct": 3, "needle_total": 3, "output_tokens": 104, "prompt_tokens": 250011, "total_s": 2314.8470665949862, "ttft_s": 1395.2731267259805, "user_number": 4, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.22106388295746185, "effective_prefill_tps": 134.55552389855143, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 2319.455895067018, "ttft_s": 1858.0508087389753, "user_number": 1, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 6.557011318432206, "effective_prefill_tps": 108.34708693301899, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 2323.056761979009, "ttft_s": 2307.5008943669964, "user_number": 2, "valid_json": null } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "2a73804cb33351a1318b0917ba061450", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "plain", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "1281f46a33da47d2b824c004", "measured_at": "2026-10-06T11:32:14.407008+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 248.79623018100392 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 140.64727534598205 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 239.41406531096436 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 48.349888990982436 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 193.9054653159692 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.6938930307588763 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 1323.9120365289455 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 102 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 94.45786571898498 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 239.77624739898602 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7294328410112476 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 677.6672277397699 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 107 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 181.69866411504336 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 239.89978868304752 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.8384524490583956 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 352.29207827015796 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 140.64727534598205 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 240.11136761197122 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0958728674515994 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 455.11724164252456 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 110 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 239.41406531096436 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 248.78772492695134 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.988237702202378 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 267.36524404637186 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 104 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 2.624256629358079 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 2.1342767115630634 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 114.19500350952148 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": true, "name": "first-token-300-s-reference", "observed": 239.41406531096436 } ], "reasons": [ "Observed decode overlap was 4 of 5 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "ae06db6e13f7e0df59a2831f51932b997a654b4e3402352238c514019f5351e2", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.6938930307588763, "effective_prefill_tps": 1323.9120365289455, "needle_correct": 3, "needle_total": 3, "output_tokens": 102, "prompt_tokens": 64011, "total_s": 193.9054653159692, "ttft_s": 48.349888990982436, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.7294328410112476, "effective_prefill_tps": 677.6672277397699, "needle_correct": 3, "needle_total": 3, "output_tokens": 107, "prompt_tokens": 64011, "total_s": 239.77624739898602, "ttft_s": 94.45786571898498, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 1.8384524490583956, "effective_prefill_tps": 352.29207827015796, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 64011, "total_s": 239.89978868304752, "ttft_s": 181.69866411504336, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 1.0958728674515994, "effective_prefill_tps": 455.11724164252456, "needle_correct": 3, "needle_total": 3, "output_tokens": 110, "prompt_tokens": 64011, "total_s": 240.11136761197122, "ttft_s": 140.64727534598205, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 10.988237702202378, "effective_prefill_tps": 267.36524404637186, "needle_correct": 3, "needle_total": 3, "output_tokens": 104, "prompt_tokens": 64011, "total_s": 248.78772492695134, "ttft_s": 239.41406531096436, "user_number": 4, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 64000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "4b077fafed0bf4c9ab19a70af78278b9", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 64000, "document": "distinct", "input_tokens": { "max": 64011, "median": 64011, "min": 64011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "143509cbac4748feb0653051", "measured_at": "2026-10-06T06:21:00.062259+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 2313.696399460023 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 2287.0217646899982 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 2288.0903925769962 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 2272.897360128991 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 2313.0252226149896 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.541874739418013 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 109.99660802360712 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 2282.2568066649837 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 2313.2767032029806 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.288212127821199 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 109.5455162056614 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 2287.0217646899982 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 2313.4932939930004 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.853196346628604 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 109.3172806048431 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 2288.0903925769962 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 2313.6827091629966 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.985571202874088 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 109.26622515049388 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 2287.0219485620037 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 2313.6792530190432 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.863856533806444 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 109.31727181595167 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 104 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 12.528959114645906 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.2230197532054878 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 104.5239028930664 } ], "model": { "artifact_sha256": [ "1d8268aa85ace093a561e3e7b63b9d390dac1cd55a90cd55b5ec509c3c9da9fe", "f078f51aed0c1dba613c1759487004ee06aba1a290606b6be125f59ab710959e" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 2288.0903925769962 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 2.541874739418013, "effective_prefill_tps": 109.99660802360712, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 2313.0252226149896, "ttft_s": 2272.897360128991, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.288212127821199, "effective_prefill_tps": 109.5455162056614, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 2313.2767032029806, "ttft_s": 2282.2568066649837, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.853196346628604, "effective_prefill_tps": 109.3172806048431, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 2313.4932939930004, "ttft_s": 2287.0217646899982, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.985571202874088, "effective_prefill_tps": 109.26622515049388, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 2313.6827091629966, "ttft_s": 2288.0903925769962, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.863856533806444, "effective_prefill_tps": 109.31727181595167, "needle_correct": 3, "needle_total": 3, "output_tokens": 104, "prompt_tokens": 250011, "total_s": 2313.6792530190432, "ttft_s": 2287.0219485620037, "user_number": 4, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "93576827588b5c19488481a382f85779", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "plain", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "14d21ed0bf0340c08e5b3228", "measured_at": "2026-10-06T09:52:10.747745+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 1108.3707114130375 }, { "basis": null, "n": 3, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 732.4510988649563 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": 3, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 1088.7835062469821 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 732.4510988649563 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 1106.5980599549948 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.2512381758390893 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 341.3347326359805 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 95 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 1088.7835062469821 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 1107.8617047460284 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.346416749207168 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 229.62416179666752 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 367.63189487101045 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 1108.3632264330518 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.14310181773527664 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 680.0579696375919 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 107 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.40770517802967926 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.27517868963820075 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 114.21642303466797 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "3/3" }, { "met": true, "name": "hidden-facts", "observed": "9/9" }, { "met": false, "name": "first-token-300-s-reference", "observed": 1088.7835062469821 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 9, "skipped": null, "suite": "hidden-facts", "total": 9, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.2512381758390893, "effective_prefill_tps": 341.3347326359805, "needle_correct": 3, "needle_total": 3, "output_tokens": 95, "prompt_tokens": 250011, "total_s": 1106.5980599549948, "ttft_s": 732.4510988649563, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.346416749207168, "effective_prefill_tps": 229.62416179666752, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1107.8617047460284, "ttft_s": 1088.7835062469821, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.14310181773527664, "effective_prefill_tps": 680.0579696375919, "needle_correct": 3, "needle_total": 3, "output_tokens": 107, "prompt_tokens": 250011, "total_s": 1108.3632264330518, "ttft_s": 367.63189487101045, "user_number": 2, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 3, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "4b077fafed0bf4c9ab19a70af78278b9", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 3, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "1572c5a898984fc582ab7e80", "measured_at": "2026-10-06T08:37:27.672763+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 1098.690269379993 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 368.0232669090037 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 1093.7382014360046 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 368.0232669090037 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 733.0965154250152 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.2821351616934008 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 104 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249984 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 368.0222821239731 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 733.4305818309658 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.30103311853672976 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 111 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 723.5712825980154 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 1094.2885796570336 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.2994195331067296 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 368.02321318001486 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 1094.5619828170165 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.1417680711677115 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 104 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249957 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 1093.7382014360046 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 1098.6503881369717 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 21.37540903714571 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 106 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.7281407667118508 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.4887637717070499 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 116.24874877929688 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 1093.7382014360046 } ], "reasons": [ "Observed decode overlap was 4 of 5 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 249984, "checks_passed": true, "completed": true, "decode_tps": 0.2821351616934008, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 104, "prompt_tokens": 250011, "total_s": 733.0965154250152, "ttft_s": 368.0232669090037, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.30103311853672976, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 111, "prompt_tokens": 250011, "total_s": 733.4305818309658, "ttft_s": 368.0222821239731, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.2994195331067296, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "total_s": 1094.2885796570336, "ttft_s": 723.5712825980154, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": 249957, "checks_passed": true, "completed": true, "decode_tps": 0.1417680711677115, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 104, "prompt_tokens": 250011, "total_s": 1094.5619828170165, "ttft_s": 368.02321318001486, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 21.37540903714571, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 106, "prompt_tokens": 250011, "total_s": 1098.6503881369717, "ttft_s": 1093.7382014360046, "user_number": 2, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "warm", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "10597aa97c3a6a4e114a3fa137d2bc6c", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp CUDA", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "1698e65ee74543aea36ae56d", "measured_at": "2026-10-06T16:09:24.389242+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 118.33478068903787 }, { "basis": null, "n": 1, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 115.21545541897649 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 1, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 115.21545541897649 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 115.21545541897649 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 118.32117482397007 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 30.588726028260243 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 2169.98664884695 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 96 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[0].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 115.34221304795938 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 30.588705653314744 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.811257682999138 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 34.96198272705078 } ], "model": { "artifact_sha256": [], "publisher_repo": "nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16", "publisher_revision": "a9904d24bcc1d289a1950fa9d2b978c47cf903b9", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/1" }, { "met": false, "name": "hidden-facts", "observed": "0/3" }, { "met": true, "name": "first-token-300-s-reference", "observed": 115.21545541897649 } ], "reasons": [ "One or more requests did not meet completion, exact retrieval, synthesis-presence, reasoning-off or token-accounting requirements." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 3, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "04cf656671acc024178598a26462c2de38b6f97137ea5abf44a09b83b8047c5e", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 0, "checks_passed": false, "completed": true, "decode_tps": 30.588726028260243, "effective_prefill_tps": 2169.98664884695, "first_real_token_from_dispatch_s": 115.34221304795938, "first_token_within_300_seconds": true, "needle_correct": 0, "needle_total": 3, "output_tokens": 96, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": false, "total_s": 118.32117482397007, "ttft_s": 115.21545541897649, "user_number": 0, "valid_json": false } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "q8_0", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "9e5118825b2267598d5691416d3ab69e", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250016, "median": 250016, "min": 250016 }, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "19df494dbc8044c3bdab1cf2", "measured_at": "2026-10-06T02:26:18.982638+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 469.25239878101274 }, { "basis": null, "n": 1, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 457.97625302005326 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 1, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 457.97625302005326 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 457.97625302005326 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 469.24367845000234 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.052644779780987 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 545.9038505847002 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 9.052642778146597 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.21949807878993344 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 103.3042221069336 } ], "model": { "artifact_sha256": [ "1d8268aa85ace093a561e3e7b63b9d390dac1cd55a90cd55b5ec509c3c9da9fe", "f078f51aed0c1dba613c1759487004ee06aba1a290606b6be125f59ab710959e" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "1/1" }, { "met": true, "name": "hidden-facts", "observed": "3/3" }, { "met": false, "name": "first-token-300-s-reference", "observed": 457.97625302005326 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 3, "skipped": null, "suite": "hidden-facts", "total": 3, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 9.052644779780987, "effective_prefill_tps": 545.9038505847002, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 469.24367845000234, "ttft_s": 457.97625302005326, "user_number": 0, "valid_json": null } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "2a73804cb33351a1318b0917ba061450", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "plain", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "20d9c54b7ed14c7797d00629", "measured_at": "2026-10-06T11:41:32.580860+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 27.06127781403484 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 10.025524195982143 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 10.418843402992934 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 10.020745349989738 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 17.6927117149462 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 14.859293507948918 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 115 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249984 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 10.025524195982143 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 19.676395021029748 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.60517035512674 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 113 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249984 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 10.022823013016023 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 22.73083613597555 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.049408344741938 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 116 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249920 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 10.026069746993016 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 23.340029031969607 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 8.03667772371351 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249984 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 10.418843402992934 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 27.048698938975576 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.456468862985403 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 125 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249957 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 33.59204146169301 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 21.321979101103256 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 114.14016723632812 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": true, "name": "first-token-300-s-reference", "observed": 10.418843402992934 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "77fda79ee1961c35a8f8a77e7d5a01d5f9ce414975e00e640a166e71e5f9093e", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 249984, "checks_passed": true, "completed": true, "decode_tps": 14.859293507948918, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 115, "prompt_tokens": 250011, "total_s": 17.6927117149462, "ttft_s": 10.020745349989738, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": 249984, "checks_passed": true, "completed": true, "decode_tps": 11.60517035512674, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 113, "prompt_tokens": 250011, "total_s": 19.676395021029748, "ttft_s": 10.025524195982143, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": 249920, "checks_passed": true, "completed": true, "decode_tps": 9.049408344741938, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 116, "prompt_tokens": 250011, "total_s": 22.73083613597555, "ttft_s": 10.022823013016023, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": 249984, "checks_passed": true, "completed": true, "decode_tps": 8.03667772371351, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "total_s": 23.340029031969607, "ttft_s": 10.026069746993016, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": 249957, "checks_passed": true, "completed": true, "decode_tps": 7.456468862985403, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 125, "prompt_tokens": 250011, "total_s": 27.048698938975576, "ttft_s": 10.418843402992934, "user_number": 4, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "warm", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "shared-record-groups-v2", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "shared", "thinking": false }, "setup_complete": false, "setup_id": "4b077fafed0bf4c9ab19a70af78278b9", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": 250000, "document": "shared", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp CUDA", "recipe": "with multi-token prediction", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "23cac8cdc0f4402ba860d186", "measured_at": "2026-10-06T13:19:49.255552+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 2309.1146278069937 }, { "basis": null, "n": 4, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 1437.8453576315078 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": 4, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 2300.6651444099844 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 568.7991483289516 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 664.4976784129976 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0240491668360288 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 439.5417973717008 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 99 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 1146.7423342610127 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 1258.9642769480124 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9623786348203294 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 218.0184619774353 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 1728.9483810020029 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 1833.8240511080367 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9725801980275657 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 144.6029290100075 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 2300.6651444099844 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 2309.10665944597 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.675449791165159 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 108.66900844195492 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.23846359091744862 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.1814548290302641 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 82.89003372192383 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "IQ3_S GGUF variant", "weights_dtype": "IQ3_S", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "4/4" }, { "met": true, "name": "hidden-facts", "observed": "12/12" }, { "met": false, "name": "first-token-300-s-reference", "observed": 2300.6651444099844 } ], "reasons": [ "Observed decode overlap was 1 of 4 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 12, "skipped": null, "suite": "hidden-facts", "total": 12, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 1.0240491668360288, "effective_prefill_tps": 439.5417973717008, "needle_correct": 3, "needle_total": 3, "output_tokens": 99, "prompt_tokens": 250011, "total_s": 664.4976784129976, "ttft_s": 568.7991483289516, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.9623786348203294, "effective_prefill_tps": 218.0184619774353, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 1258.9642769480124, "ttft_s": 1146.7423342610127, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.9725801980275657, "effective_prefill_tps": 144.6029290100075, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1833.8240511080367, "ttft_s": 1728.9483810020029, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 12.675449791165159, "effective_prefill_tps": 108.66900844195492, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "total_s": 2309.10665944597, "ttft_s": 2300.6651444099844, "user_number": 1, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 4, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "q8_0", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "fa4cd389621fd181fb8cb62d87434f73", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011.0, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 4, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "275ce298de56481bba5b7b7b", "measured_at": "2026-10-06T08:49:11.110911+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 631.9320386610343 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 377.18796812195797 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 626.958289237984 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 377.18796812195797 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 503.0469450709643 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8263216698649727 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 339.3825116887334 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 105 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 131.22718459897442 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 627.912521288963 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.22952157347692007 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 975.4914760321731 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 115 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 496.14283797499957 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 628.1510672860313 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8863083810050202 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 258.0123911945907 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 118 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 256.44749687000876 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 628.3502307659946 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.2957762607649045 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 499.17040159252474 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 111 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 626.958289237984 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 631.8937066419749 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 20.66694499183006 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 204.17785711962878 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 1.0925419492885327 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.8735116535151504 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 116.34001922607422 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 626.958289237984 } ], "reasons": [ "Observed decode overlap was 4 of 5 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "82b4b7218007663e7b3bff534276ade2784e677bfbd1b1499218ec45d28b1b59", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.8263216698649727, "effective_prefill_tps": 339.3825116887334, "needle_correct": 3, "needle_total": 3, "output_tokens": 105, "prompt_tokens": 128011, "total_s": 503.0469450709643, "ttft_s": 377.18796812195797, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.22952157347692007, "effective_prefill_tps": 975.4914760321731, "needle_correct": 3, "needle_total": 3, "output_tokens": 115, "prompt_tokens": 128011, "total_s": 627.912521288963, "ttft_s": 131.22718459897442, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.8863083810050202, "effective_prefill_tps": 258.0123911945907, "needle_correct": 3, "needle_total": 3, "output_tokens": 118, "prompt_tokens": 128011, "total_s": 628.1510672860313, "ttft_s": 496.14283797499957, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.2957762607649045, "effective_prefill_tps": 499.17040159252474, "needle_correct": 3, "needle_total": 3, "output_tokens": 111, "prompt_tokens": 128011, "total_s": 628.3502307659946, "ttft_s": 256.44749687000876, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 20.66694499183006, "effective_prefill_tps": 204.17785711962878, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 128011, "total_s": 631.8937066419749, "ttft_s": 626.958289237984, "user_number": 1, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 128000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "10597aa97c3a6a4e114a3fa137d2bc6c", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 128000, "document": "distinct", "input_tokens": { "max": 128011, "median": 128011, "min": 128011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "28597e232406401b9487303b", "measured_at": "2026-10-06T04:49:28.129613+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 482.07286677503726 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 467.2595032850513 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 467.2605302440352 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 467.2595032850513 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 482.0336123970337 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.310085446195036 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 535.0581384483495 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 467.2605302440352 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 482.03091158304596 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.311930377501891 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 535.0569624817814 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 467.2598849820206 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 482.03054342401447 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.311793202999636 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 535.0577013680942 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 467.25898227799917 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 482.0360517700319 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.308620972394398 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 535.0587350533887 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 467.25858435896225 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 482.0393273359514 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.306804547520779 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 535.0591907112699 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 36.53402061100703 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 1.1305344846432275 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 103.58668899536133 } ], "model": { "artifact_sha256": [ "1d8268aa85ace093a561e3e7b63b9d390dac1cd55a90cd55b5ec509c3c9da9fe", "f078f51aed0c1dba613c1759487004ee06aba1a290606b6be125f59ab710959e" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 467.2605302440352 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "2d23afb10cda3f1d9b23f9375d8c3456fc82a4da61760dec43722fa7b44f92f8", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.310085446195036, "effective_prefill_tps": 535.0581384483495, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 482.0336123970337, "ttft_s": 467.2595032850513, "user_number": 2, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.311930377501891, "effective_prefill_tps": 535.0569624817814, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 482.03091158304596, "ttft_s": 467.2605302440352, "user_number": 1, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.311793202999636, "effective_prefill_tps": 535.0577013680942, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 482.03054342401447, "ttft_s": 467.2598849820206, "user_number": 3, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.308620972394398, "effective_prefill_tps": 535.0587350533887, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 482.0360517700319, "ttft_s": 467.25898227799917, "user_number": 0, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.306804547520779, "effective_prefill_tps": 535.0591907112699, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 482.0393273359514, "ttft_s": 467.25858435896225, "user_number": 4, "valid_json": null } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "shared-question-order-v1", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "shared", "thinking": false }, "setup_complete": false, "setup_id": "2a73804cb33351a1318b0917ba061450", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "shared", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "2f6f82fd67524b1daac0f59e", "measured_at": "2026-10-06T02:10:27.915104+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 66.34113659203285 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 66.32670700299786 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 66.33214250701712 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.0 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 103.64082717895508 } ], "model": { "artifact_sha256": [ "1d8268aa85ace093a561e3e7b63b9d390dac1cd55a90cd55b5ec509c3c9da9fe", "f078f51aed0c1dba613c1759487004ee06aba1a290606b6be125f59ab710959e" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/2" }, { "met": false, "name": "hidden-facts", "observed": "0/6" }, { "met": false, "name": "first-token-300-s-reference", "observed": null } ], "reasons": [ "Observed decode overlap was 0 of 2 requested users.", "One or more requests did not meet completion, exact retrieval, synthesis-presence, reasoning-off or token-accounting requirements.", "The two-user attempt was intentionally cancelled after the relay limitation was diagnosed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 6, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": false, "completed": false, "decode_tps": null, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": null, "prompt_tokens": null, "total_s": 66.32670700299786, "ttft_s": null, "user_number": 1, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": false, "decode_tps": null, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": null, "prompt_tokens": null, "total_s": 66.33214250701712, "ttft_s": null, "user_number": 0, "valid_json": null } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 2, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "2a73804cb33351a1318b0917ba061450", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "37c3ae60d5ac49b28c7b937a", "measured_at": "2026-10-06T02:08:42.396256+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 300.5277418959886 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 300.49065901502036 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.0 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 103.57308197021484 } ], "model": { "artifact_sha256": [ "1d8268aa85ace093a561e3e7b63b9d390dac1cd55a90cd55b5ec509c3c9da9fe", "f078f51aed0c1dba613c1759487004ee06aba1a290606b6be125f59ab710959e" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/1" }, { "met": false, "name": "hidden-facts", "observed": "0/3" }, { "met": false, "name": "first-token-300-s-reference", "observed": null } ], "reasons": [ "Observed decode overlap was 0 of 1 requested users.", "One or more requests did not meet completion, exact retrieval, synthesis-presence, reasoning-off or token-accounting requirements.", "The stream closed at the 300-second relay idle limit before the first token; the corrected recipe extends that limit." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 3, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": false, "completed": false, "decode_tps": null, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": null, "prompt_tokens": null, "total_s": 300.49065901502036, "ttft_s": null, "user_number": 0, "valid_json": null } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "2a73804cb33351a1318b0917ba061450", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "3b0fd2f8a514494a9a4b58eb", "measured_at": "2026-10-06T02:42:35.910138+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 933.1101417099708 }, { "basis": null, "n": 2, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 695.598551693518 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": 2, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 919.2778929760098 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 471.91921041102614 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 929.4683107149904 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.22292689447370392 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 529.7750006452346 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 919.2778929760098 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 933.0668549399707 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.397221071940671 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 271.96455164458575 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.4423706057122897 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.22076707860284828 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 103.3840446472168 } ], "model": { "artifact_sha256": [ "1d8268aa85ace093a561e3e7b63b9d390dac1cd55a90cd55b5ec509c3c9da9fe", "f078f51aed0c1dba613c1759487004ee06aba1a290606b6be125f59ab710959e" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "2/2" }, { "met": true, "name": "hidden-facts", "observed": "6/6" }, { "met": false, "name": "first-token-300-s-reference", "observed": 919.2778929760098 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 6, "skipped": null, "suite": "hidden-facts", "total": 6, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.22292689447370392, "effective_prefill_tps": 529.7750006452346, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 929.4683107149904, "ttft_s": 471.91921041102614, "user_number": 1, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.397221071940671, "effective_prefill_tps": 271.96455164458575, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 933.0668549399707, "ttft_s": 919.2778929760098, "user_number": 0, "valid_json": null } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 2, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "2a73804cb33351a1318b0917ba061450", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011.0, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "3d6ef7dbb3fe4181b4ae8daa", "measured_at": "2026-10-06T06:38:46.294890+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 353.46790590899764 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 353.46063730400056 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 115.99139022827148 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/1" }, { "met": false, "name": "hidden-facts", "observed": "0/3" }, { "met": false, "name": "first-token-300-s-reference", "observed": null } ], "reasons": [ "Observed decode overlap was 0 of 1 requested users.", "One or more requests did not meet completion, exact retrieval, synthesis-presence, reasoning-off or token-accounting requirements.", "Intentionally interrupted to correct collection of retraction and Mamba gauges; this attempt does not establish an inference or memory failure." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 3, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": false, "completed": false, "decode_tps": null, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": null, "prompt_tokens": null, "total_s": 353.46063730400056, "ttft_s": null, "user_number": 0, "valid_json": false } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "10597aa97c3a6a4e114a3fa137d2bc6c", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Startup, inference completion, concurrent decode and retrieval correctness are separate gates.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "43fdf55445c8479cba2e768c", "measured_at": "2026-10-06T01:58:43.887855+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/0" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "did_not_start" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "long-context-concurrency", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "2cdb4314c031e68844e1f2572dc8372d", "standard": "pre-standard", "system": { "driver": null, "hardware": "unrecorded hardware", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "plain", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "449efa9bb9be4eaabacc10d6", "measured_at": "2026-10-06T11:15:16.518028+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 1475.3738709769677 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 726.1311567630037 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 1462.3550099330023 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 364.9085966430139 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 1102.353035215987 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.13831550509402438 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 726.1311567630037 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 1102.352675920003 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.2711168681380897 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 1079.2620187569992 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 1102.3486223660293 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.418146632885556 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 364.9092461949913 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 1462.8138431169791 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.0974583768935644 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249984 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 1462.3550099330023 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 1475.3367804040317 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 8.55045159269387 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.4718900740835121 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.35855318465800473 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 114.22733688354492 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 1462.3550099330023 } ], "reasons": [ "Observed decode overlap was 4 of 5 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.13831550509402438, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1102.353035215987, "ttft_s": 364.9085966430139, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.2711168681380897, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1102.352675920003, "ttft_s": 726.1311567630037, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 4.418146632885556, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1102.3486223660293, "ttft_s": 1079.2620187569992, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": 249984, "checks_passed": true, "completed": true, "decode_tps": 0.0974583768935644, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "total_s": 1462.8138431169791, "ttft_s": 364.9092461949913, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 8.55045159269387, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "total_s": 1475.3367804040317, "ttft_s": 1462.3550099330023, "user_number": 0, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "warm", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "4b077fafed0bf4c9ab19a70af78278b9", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp CUDA", "recipe": "with multi-token prediction", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "48e8f2715643437ea845c968", "measured_at": "2026-10-06T14:08:48.934522+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 42.49571285705315 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 3.4209511509980075 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 3.4245078209787607 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 3.419261311995797 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 40.740408156008925 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.8938017486813865 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 250007 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 3.423158894001972 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 40.75499000196578 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.9733339272586856 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 250007 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 3.4209511509980075 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 41.594886083039455 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.802959668435677 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 250007 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 3.4200159200117923 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 42.18678043503314 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.8116867983079854 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 110 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 250007 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 3.4245078209787607 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 42.48903142201016 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.8414528008494475 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 250007 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 13.975137973739782 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 12.966013815404176 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 85.46923065185547 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "IQ3_S GGUF variant", "weights_dtype": "IQ3_S", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": true, "name": "first-token-300-s-reference", "observed": 3.4245078209787607 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 250007, "checks_passed": true, "completed": true, "decode_tps": 2.8938017486813865, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 40.740408156008925, "ttft_s": 3.419261311995797, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": 250007, "checks_passed": true, "completed": true, "decode_tps": 2.9733339272586856, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "total_s": 40.75499000196578, "ttft_s": 3.423158894001972, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": 250007, "checks_passed": true, "completed": true, "decode_tps": 2.802959668435677, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "total_s": 41.594886083039455, "ttft_s": 3.4209511509980075, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": 250007, "checks_passed": true, "completed": true, "decode_tps": 2.8116867983079854, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 110, "prompt_tokens": 250011, "total_s": 42.18678043503314, "ttft_s": 3.4200159200117923, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": 250007, "checks_passed": true, "completed": true, "decode_tps": 2.8414528008494475, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "total_s": 42.48903142201016, "ttft_s": 3.4245078209787607, "user_number": 3, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "warm", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "q8_0", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "fa4cd389621fd181fb8cb62d87434f73", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp CUDA", "recipe": "with multi-token prediction", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "4bfc3fa5a5e04f06a8295055", "measured_at": "2026-10-06T14:28:19.890302+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 1169.3848097359878 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 700.4697799090063 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 1163.4144326859969 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 229.71511798800202 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 345.57731081097154 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9752965764479982 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 557.2597969224037 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 114 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 465.2950279919896 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 560.1778816450387 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0960884500827608 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 275.11791938211684 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 105 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 700.4697799090063 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 795.8830416710116 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.1004759512562046 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 182.75021088936788 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 106 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 934.1679784459993 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 1032.926016065001 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.063204601179695 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 137.03210017212106 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 106 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 1163.4144326859969 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 1169.3443901579594 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 18.718515356108266 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 110.0304383404103 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.5725666276780521 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.46434671929986265 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 80.6306266784668 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "IQ3_S GGUF variant", "weights_dtype": "IQ3_S", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 1163.4144326859969 } ], "reasons": [ "Observed decode overlap was 1 of 5 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "82b4b7218007663e7b3bff534276ade2784e677bfbd1b1499218ec45d28b1b59", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.9752965764479982, "effective_prefill_tps": 557.2597969224037, "needle_correct": 3, "needle_total": 3, "output_tokens": 114, "prompt_tokens": 128011, "total_s": 345.57731081097154, "ttft_s": 229.71511798800202, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 1.0960884500827608, "effective_prefill_tps": 275.11791938211684, "needle_correct": 3, "needle_total": 3, "output_tokens": 105, "prompt_tokens": 128011, "total_s": 560.1778816450387, "ttft_s": 465.2950279919896, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 1.1004759512562046, "effective_prefill_tps": 182.75021088936788, "needle_correct": 3, "needle_total": 3, "output_tokens": 106, "prompt_tokens": 128011, "total_s": 795.8830416710116, "ttft_s": 700.4697799090063, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 1.063204601179695, "effective_prefill_tps": 137.03210017212106, "needle_correct": 3, "needle_total": 3, "output_tokens": 106, "prompt_tokens": 128011, "total_s": 1032.926016065001, "ttft_s": 934.1679784459993, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 18.718515356108266, "effective_prefill_tps": 110.0304383404103, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 128011, "total_s": 1169.3443901579594, "ttft_s": 1163.4144326859969, "user_number": 4, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 128000, "fixture_variant": "distinct-v1", "kv_dtype": "q8_0", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "fa4cd389621fd181fb8cb62d87434f73", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 128000, "document": "distinct", "input_tokens": { "max": 128011, "median": 128011, "min": 128011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "plain", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "4c073dc41eb44867a15e1609", "measured_at": "2026-10-06T11:27:04.162387+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 643.9211764599895 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 372.4602032719995 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 632.8362723499886 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 493.1619223859743 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 509.61548347899225 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.381597236391385 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 259.5719462294818 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 106 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 252.03524400299648 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 511.7796049779863 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.41194349551366305 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 507.90912400520483 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 372.4602032719995 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 633.1171602199902 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.41817414457788266 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 343.6904100772248 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 110 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 127.46482677699532 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 633.7299071159796 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.22517847749575784 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 1004.2848936197923 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 115 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 632.8362723499886 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 643.8795158120338 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.96084170181176 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 202.28138871471612 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 111 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 1.0553445281353664 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.8541418113062704 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 114.19036483764648 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 632.8362723499886 } ], "reasons": [ "Observed decode overlap was 4 of 5 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "82b4b7218007663e7b3bff534276ade2784e677bfbd1b1499218ec45d28b1b59", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 6.381597236391385, "effective_prefill_tps": 259.5719462294818, "needle_correct": 3, "needle_total": 3, "output_tokens": 106, "prompt_tokens": 128011, "total_s": 509.61548347899225, "ttft_s": 493.1619223859743, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.41194349551366305, "effective_prefill_tps": 507.90912400520483, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 128011, "total_s": 511.7796049779863, "ttft_s": 252.03524400299648, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.41817414457788266, "effective_prefill_tps": 343.6904100772248, "needle_correct": 3, "needle_total": 3, "output_tokens": 110, "prompt_tokens": 128011, "total_s": 633.1171602199902, "ttft_s": 372.4602032719995, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.22517847749575784, "effective_prefill_tps": 1004.2848936197923, "needle_correct": 3, "needle_total": 3, "output_tokens": 115, "prompt_tokens": 128011, "total_s": 633.7299071159796, "ttft_s": 127.46482677699532, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 9.96084170181176, "effective_prefill_tps": 202.28138871471612, "needle_correct": 3, "needle_total": 3, "output_tokens": 111, "prompt_tokens": 128011, "total_s": 643.8795158120338, "ttft_s": 632.8362723499886, "user_number": 0, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 128000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "4b077fafed0bf4c9ab19a70af78278b9", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 128000, "document": "distinct", "input_tokens": { "max": 128011, "median": 128011, "min": 128011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "5c550512184f4a06a45faa77", "measured_at": "2026-10-06T05:27:19.850286+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 26.837993788009044 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 10.883544983051252 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 10.885174758033827 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 8.128231940965634 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 25.007480574015062 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.161411699118458 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 105 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 10.885174758033827 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 26.020307878032327 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.333929547889633 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 10.883544983051252 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 26.143479022022802 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.339481266037523 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 113 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 3.581799208011944 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 26.618511073000263 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.122258796809405 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 119 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 10.884514423029032 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 26.82741090201307 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.401415430097527 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 119 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 24.21928851925985 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 21.164026062699847 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 103.37997436523438 } ], "model": { "artifact_sha256": [ "1d8268aa85ace093a561e3e7b63b9d390dac1cd55a90cd55b5ec509c3c9da9fe", "f078f51aed0c1dba613c1759487004ee06aba1a290606b6be125f59ab710959e" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": true, "name": "first-token-300-s-reference", "observed": 10.885174758033827 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "77fda79ee1961c35a8f8a77e7d5a01d5f9ce414975e00e640a166e71e5f9093e", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 6.161411699118458, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 105, "prompt_tokens": 250011, "total_s": 25.007480574015062, "ttft_s": 8.128231940965634, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.333929547889633, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "total_s": 26.020307878032327, "ttft_s": 10.885174758033827, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.339481266037523, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 113, "prompt_tokens": 250011, "total_s": 26.143479022022802, "ttft_s": 10.883544983051252, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.122258796809405, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 119, "prompt_tokens": 250011, "total_s": 26.618511073000263, "ttft_s": 3.581799208011944, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.401415430097527, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 119, "prompt_tokens": 250011, "total_s": 26.82741090201307, "ttft_s": 10.884514423029032, "user_number": 3, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "warm", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "shared-record-groups-v2", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "shared", "thinking": false }, "setup_complete": false, "setup_id": "2a73804cb33351a1318b0917ba061450", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": 250000, "document": "shared", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "plain", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "5f884458af3a445386be7e43", "measured_at": "2026-10-06T09:32:39.922345+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 735.7327210759977 }, { "basis": null, "n": 2, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 542.3895939379872 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": 2, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 719.9948586159735 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 364.78432926000096 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 733.4575591139728 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.23869376150488605 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 685.3666123957973 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 89 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 719.9948586159735 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 735.6937309199711 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.8157761862139 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 347.23997957511716 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.5257350942722262 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.2677602808148733 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 114.12946319580078 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "2/2" }, { "met": true, "name": "hidden-facts", "observed": "6/6" }, { "met": false, "name": "first-token-300-s-reference", "observed": 719.9948586159735 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 6, "skipped": null, "suite": "hidden-facts", "total": 6, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.23869376150488605, "effective_prefill_tps": 685.3666123957973, "needle_correct": 3, "needle_total": 3, "output_tokens": 89, "prompt_tokens": 250011, "total_s": 733.4575591139728, "ttft_s": 364.78432926000096, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 6.8157761862139, "effective_prefill_tps": 347.23997957511716, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "total_s": 735.6937309199711, "ttft_s": 719.9948586159735, "user_number": 1, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 2, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "4b077fafed0bf4c9ab19a70af78278b9", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011.0, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "60aa08187a1b4b7d8ca92f35", "measured_at": "2026-10-06T03:38:09.788940+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 1858.4793928060099 }, { "basis": null, "n": 4, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 1162.9137839740142 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": 4, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 1842.9074569240329 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 471.9631636390113 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 1844.1237283250084 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.07433532388634234 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 529.7256634867906 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 932.129960403021 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 1850.0708964529913 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.11111826043941392 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 268.21474539012115 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 1393.6976075450075 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 1854.8268619989976 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.22119611587162316 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 179.38683301637673 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 1842.9074569240329 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 1858.4381864540046 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.5676245152011035 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 135.66117987133728 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.29427152647481986 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.22168661196611128 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 103.45914840698242 } ], "model": { "artifact_sha256": [ "1d8268aa85ace093a561e3e7b63b9d390dac1cd55a90cd55b5ec509c3c9da9fe", "f078f51aed0c1dba613c1759487004ee06aba1a290606b6be125f59ab710959e" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "4/4" }, { "met": true, "name": "hidden-facts", "observed": "12/12" }, { "met": false, "name": "first-token-300-s-reference", "observed": 1842.9074569240329 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 12, "skipped": null, "suite": "hidden-facts", "total": 12, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.07433532388634234, "effective_prefill_tps": 529.7256634867906, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1844.1237283250084, "ttft_s": 471.9631636390113, "user_number": 2, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.11111826043941392, "effective_prefill_tps": 268.21474539012115, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1850.0708964529913, "ttft_s": 932.129960403021, "user_number": 0, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.22119611587162316, "effective_prefill_tps": 179.38683301637673, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1854.8268619989976, "ttft_s": 1393.6976075450075, "user_number": 1, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 6.5676245152011035, "effective_prefill_tps": 135.66117987133728, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1858.4381864540046, "ttft_s": 1842.9074569240329, "user_number": 3, "valid_json": null } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 4, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "2a73804cb33351a1318b0917ba061450", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011.0, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 4, "warmup": null } }, { "caveats": [ "Startup, inference completion, concurrent decode and retrieval correctness are separate gates.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "616783b949f34ed6baaf3123", "measured_at": "2026-10-06T02:16:57.570647+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/0" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "did_not_start" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "long-context-concurrency", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "2cdb4314c031e68844e1f2572dc8372d", "standard": "pre-standard", "system": { "driver": null, "hardware": "unrecorded hardware", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "641d1e53d16a453a8834e31f", "measured_at": "2026-10-06T06:22:47.697268+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 36.32477205700707 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 8.837211286998354 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 10.696028016973287 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 4.630959067028016 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 35.70422672299901 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.2825643292265667 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 8.836861174961086 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 35.967839126009494 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.759540116247762 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 8.837211286998354 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 35.967960021982435 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.7595718789903807 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 10.696028016973287 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 36.17400689300848 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.003457279570243 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 10.695194705971517 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 36.31433553196257 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.020431469563755 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 104 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 16.12842033907988 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 14.205182050150356 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 104.31328201293945 } ], "model": { "artifact_sha256": [ "1d8268aa85ace093a561e3e7b63b9d390dac1cd55a90cd55b5ec509c3c9da9fe", "f078f51aed0c1dba613c1759487004ee06aba1a290606b6be125f59ab710959e" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": true, "name": "first-token-300-s-reference", "observed": 10.696028016973287 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.2825643292265667, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 35.70422672299901, "ttft_s": 4.630959067028016, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.759540116247762, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 35.967839126009494, "ttft_s": 8.836861174961086, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.7595718789903807, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 35.967960021982435, "ttft_s": 8.837211286998354, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 4.003457279570243, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 36.17400689300848, "ttft_s": 10.696028016973287, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 4.020431469563755, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 104, "prompt_tokens": 250011, "total_s": 36.31433553196257, "ttft_s": 10.695194705971517, "user_number": 4, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "warm", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "93576827588b5c19488481a382f85779", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "plain", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "6963b3247ffd47cf944231e9", "measured_at": "2026-10-06T09:19:16.406605+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 370.6741369399824 }, { "basis": null, "n": 1, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 360.526133254054 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 1, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 360.526133254054 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 360.526133254054 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 370.66455062804744 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 8.679855716507907 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 693.4615189845983 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 89 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 8.679852766059257 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.24010307472412193 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 114.08170700073242 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "1/1" }, { "met": true, "name": "hidden-facts", "observed": "3/3" }, { "met": false, "name": "first-token-300-s-reference", "observed": 360.526133254054 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 3, "skipped": null, "suite": "hidden-facts", "total": 3, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 8.679855716507907, "effective_prefill_tps": 693.4615189845983, "needle_correct": 3, "needle_total": 3, "output_tokens": 89, "prompt_tokens": 250011, "total_s": 370.66455062804744, "ttft_s": 360.526133254054, "user_number": 0, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "4b077fafed0bf4c9ab19a70af78278b9", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "6b3b5039f1b44dbcb2d4321a", "measured_at": "2026-10-06T09:01:43.897039+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 391.2430079850019 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 373.2503249149886 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 384.64660948503297 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 373.2503209949937 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 383.1554266430321 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.30729988954534 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 113 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249664 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 373.2503249149886 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 385.57688139303355 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.653953252228474 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 120 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249664 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 373.25073302700184 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 385.5769596970058 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.329700246374184 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 116 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 245760 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 373.22135591396363 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 385.8861167799914 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.790946809948863 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 669.8732428849368 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 125 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 384.64660948503297 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 391.20350280601997 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.318703807109284 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249664 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 32.08611573542737 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 1.4875665203512358 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 116.11288833618164 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 384.64660948503297 } ], "reasons": [ "Observed decode overlap was 4 of 5 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "77fda79ee1961c35a8f8a77e7d5a01d5f9ce414975e00e640a166e71e5f9093e", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 249664, "checks_passed": true, "completed": true, "decode_tps": 11.30729988954534, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 113, "prompt_tokens": 250011, "total_s": 383.1554266430321, "ttft_s": 373.2503209949937, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": 249664, "checks_passed": true, "completed": true, "decode_tps": 9.653953252228474, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 120, "prompt_tokens": 250011, "total_s": 385.57688139303355, "ttft_s": 373.2503249149886, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": 245760, "checks_passed": true, "completed": true, "decode_tps": 9.329700246374184, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 116, "prompt_tokens": 250011, "total_s": 385.5769596970058, "ttft_s": 373.25073302700184, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 9.790946809948863, "effective_prefill_tps": 669.8732428849368, "needle_correct": 3, "needle_total": 3, "output_tokens": 125, "prompt_tokens": 250011, "total_s": 385.8861167799914, "ttft_s": 373.22135591396363, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": 249664, "checks_passed": true, "completed": true, "decode_tps": 16.318703807109284, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "total_s": 391.20350280601997, "ttft_s": 384.64660948503297, "user_number": 1, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "shared-record-groups-v2", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "shared", "thinking": false }, "setup_complete": false, "setup_id": "10597aa97c3a6a4e114a3fa137d2bc6c", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "shared", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp CUDA", "recipe": "with multi-token prediction", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "6d004e21988d469ba3a38aa1", "measured_at": "2026-10-06T15:25:04.252308+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 2888.3935430310084 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 1729.64179039601 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 2879.6453400250175 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 570.0103024260025 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 669.9265457559959 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.1109304783747747 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 438.6078618858927 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 1149.810656376998 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 1249.5870304310229 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.032308509670131 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 217.43666977985185 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 104 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 1729.64179039601 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 1834.618233547022 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0573800813609473 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 144.5449580301588 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 2308.15205013199 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 2412.8776892700116 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0599123663853633 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 108.31652099596441 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 2879.6453400250175 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 2888.3534509119927 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.40223067916334 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 86.82006652868856 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.2346502846316623 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.19007105223753307 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 85.52237701416016 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "IQ3_S GGUF variant", "weights_dtype": "IQ3_S", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 2879.6453400250175 } ], "reasons": [ "Observed decode overlap was 1 of 5 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "77fda79ee1961c35a8f8a77e7d5a01d5f9ce414975e00e640a166e71e5f9093e", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 1.1109304783747747, "effective_prefill_tps": 438.6078618858927, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "total_s": 669.9265457559959, "ttft_s": 570.0103024260025, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 1.032308509670131, "effective_prefill_tps": 217.43666977985185, "needle_correct": 3, "needle_total": 3, "output_tokens": 104, "prompt_tokens": 250011, "total_s": 1249.5870304310229, "ttft_s": 1149.810656376998, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 1.0573800813609473, "effective_prefill_tps": 144.5449580301588, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "total_s": 1834.618233547022, "ttft_s": 1729.64179039601, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 1.0599123663853633, "effective_prefill_tps": 108.31652099596441, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "total_s": 2412.8776892700116, "ttft_s": 2308.15205013199, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 12.40223067916334, "effective_prefill_tps": 86.82006652868856, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 2888.3534509119927, "ttft_s": 2879.6453400250175, "user_number": 2, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "shared-record-groups-v2", "kv_dtype": "q8_0", "max_output_tokens": 512, "scenario": "shared", "thinking": false }, "setup_complete": false, "setup_id": "fa4cd389621fd181fb8cb62d87434f73", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "shared", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "7f54527e97884e51b031bb15", "measured_at": "2026-10-06T05:21:11.664054+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 483.065161378996 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 467.13506713398965 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 467.1377381539787 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 467.1349509260035 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 481.32547434902517 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.328834666611245 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 535.2008011911807 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 105 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 467.13506713398965 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 482.2255145389936 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.355646722786547 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 535.2006680506574 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 467.1377381539787 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 482.46969576599076 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.370226481155185 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 535.1976078575586 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 114 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 467.1260982109816 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 483.04347071400844 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.41328381788898 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 535.2109440202597 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 119 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 467.1376734820078 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 483.0502197990427 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.41553222526538 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 535.1976819519554 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 119 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 35.42824175709795 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 1.1778949207922336 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 103.45968246459961 } ], "model": { "artifact_sha256": [ "1d8268aa85ace093a561e3e7b63b9d390dac1cd55a90cd55b5ec509c3c9da9fe", "f078f51aed0c1dba613c1759487004ee06aba1a290606b6be125f59ab710959e" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 467.1377381539787 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "77fda79ee1961c35a8f8a77e7d5a01d5f9ce414975e00e640a166e71e5f9093e", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.328834666611245, "effective_prefill_tps": 535.2008011911807, "needle_correct": 3, "needle_total": 3, "output_tokens": 105, "prompt_tokens": 250011, "total_s": 481.32547434902517, "ttft_s": 467.1349509260035, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.355646722786547, "effective_prefill_tps": 535.2006680506574, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "total_s": 482.2255145389936, "ttft_s": 467.13506713398965, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.370226481155185, "effective_prefill_tps": 535.1976078575586, "needle_correct": 3, "needle_total": 3, "output_tokens": 114, "prompt_tokens": 250011, "total_s": 482.46969576599076, "ttft_s": 467.1377381539787, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.41328381788898, "effective_prefill_tps": 535.2109440202597, "needle_correct": 3, "needle_total": 3, "output_tokens": 119, "prompt_tokens": 250011, "total_s": 483.04347071400844, "ttft_s": 467.1260982109816, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.41553222526538, "effective_prefill_tps": 535.1976819519554, "needle_correct": 3, "needle_total": 3, "output_tokens": 119, "prompt_tokens": 250011, "total_s": 483.0502197990427, "ttft_s": 467.1376734820078, "user_number": 4, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "shared-record-groups-v2", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "shared", "thinking": false }, "setup_complete": false, "setup_id": "2a73804cb33351a1318b0917ba061450", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "shared", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "8061ab66b54741d48d94625a", "measured_at": "2026-10-06T07:21:14.094581+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 1095.955539574963 }, { "basis": null, "n": 3, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 733.6205576900393 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": 3, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 1087.1728219110519 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 733.6205576900393 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 1095.4467553020222 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.28190330239543154 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 340.790613593508 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 1087.1728219110519 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 1095.9425357160508 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.657197540098506 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 229.96435797624721 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 369.61302903399337 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 1095.9425634919899 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.14869283827290877 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 676.4128435986667 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.44194815786569336 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.29563243060540095 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 116.21009826660156 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "3/3" }, { "met": true, "name": "hidden-facts", "observed": "9/9" }, { "met": false, "name": "first-token-300-s-reference", "observed": 1087.1728219110519 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 9, "skipped": null, "suite": "hidden-facts", "total": 9, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.28190330239543154, "effective_prefill_tps": 340.790613593508, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1095.4467553020222, "ttft_s": 733.6205576900393, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 12.657197540098506, "effective_prefill_tps": 229.96435797624721, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "total_s": 1095.9425357160508, "ttft_s": 1087.1728219110519, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.14869283827290877, "effective_prefill_tps": 676.4128435986667, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 1095.9425634919899, "ttft_s": 369.61302903399337, "user_number": 1, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 3, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "10597aa97c3a6a4e114a3fa137d2bc6c", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 3, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "plain", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "8092de5e5ac24e4fb69ec817", "measured_at": "2026-10-06T11:40:03.755033+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 401.9808458079933 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 373.6482511879876 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 388.36319749604445 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 373.6482511879876 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 387.1771426090272 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.539420402278763 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 245760 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 373.64766110799974 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 388.8830818479764 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.08874417341265 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249664 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 373.64846774801845 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 389.4119807080133 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.105015251628064 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 113 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249664 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 373.4947378409561 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 390.0194283769815 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.080314136287668 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 669.3829247641537 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 118 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 388.36319749604445 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 401.9723628209904 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 8.450187594473716 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 116 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249664 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 19.453967769604457 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 1.3906135225831309 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 114.28803253173828 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 388.36319749604445 } ], "reasons": [ "Observed decode overlap was 4 of 5 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "77fda79ee1961c35a8f8a77e7d5a01d5f9ce414975e00e640a166e71e5f9093e", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 245760, "checks_passed": true, "completed": true, "decode_tps": 7.539420402278763, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 387.1771426090272, "ttft_s": 373.6482511879876, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": 249664, "checks_passed": true, "completed": true, "decode_tps": 7.08874417341265, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 388.8830818479764, "ttft_s": 373.64766110799974, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": 249664, "checks_passed": true, "completed": true, "decode_tps": 7.105015251628064, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 113, "prompt_tokens": 250011, "total_s": 389.4119807080133, "ttft_s": 373.64846774801845, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.080314136287668, "effective_prefill_tps": 669.3829247641537, "needle_correct": 3, "needle_total": 3, "output_tokens": 118, "prompt_tokens": 250011, "total_s": 390.0194283769815, "ttft_s": 373.4947378409561, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": 249664, "checks_passed": true, "completed": true, "decode_tps": 8.450187594473716, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 116, "prompt_tokens": 250011, "total_s": 401.9723628209904, "ttft_s": 388.36319749604445, "user_number": 3, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "shared-record-groups-v2", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "shared", "thinking": false }, "setup_complete": false, "setup_id": "4b077fafed0bf4c9ab19a70af78278b9", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "shared", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Startup, inference completion, concurrent decode and retrieval correctness are separate gates.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "826caf17ab8f419a9f3d7b88", "measured_at": "2026-10-06T05:40:59.267029+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/0" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "did_not_start" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "long-context-concurrency", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "2cdb4314c031e68844e1f2572dc8372d", "standard": "pre-standard", "system": { "driver": null, "hardware": "unrecorded hardware", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "8a460b5cc8344abe948aa68f", "measured_at": "2026-10-06T04:33:29.956093+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 917.791812009993 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 552.7236815490178 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 903.1494155289838 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 198.05469924403587 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 909.930562952999 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.14749762613517578 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 646.3416444477767 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 106 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 375.2308839930338 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 912.2285610910039 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.19366936662004625 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 341.1526221876143 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 105 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 552.7236815490178 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 913.6971694999957 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.27425836884022 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 231.60035343744804 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 100 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 738.8299420570256 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 916.1776395619963 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.5864186649340913 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 173.26179234641745 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 105 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 903.1494155289838 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 917.7854235619889 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.174087344255268 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 141.73845190944695 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 128011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 106 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.7183239137053788 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.5687564360122189 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 103.37869262695312 } ], "model": { "artifact_sha256": [ "1d8268aa85ace093a561e3e7b63b9d390dac1cd55a90cd55b5ec509c3c9da9fe", "f078f51aed0c1dba613c1759487004ee06aba1a290606b6be125f59ab710959e" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 903.1494155289838 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "82b4b7218007663e7b3bff534276ade2784e677bfbd1b1499218ec45d28b1b59", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.14749762613517578, "effective_prefill_tps": 646.3416444477767, "needle_correct": 3, "needle_total": 3, "output_tokens": 106, "prompt_tokens": 128011, "total_s": 909.930562952999, "ttft_s": 198.05469924403587, "user_number": 4, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.19366936662004625, "effective_prefill_tps": 341.1526221876143, "needle_correct": 3, "needle_total": 3, "output_tokens": 105, "prompt_tokens": 128011, "total_s": 912.2285610910039, "ttft_s": 375.2308839930338, "user_number": 3, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.27425836884022, "effective_prefill_tps": 231.60035343744804, "needle_correct": 3, "needle_total": 3, "output_tokens": 100, "prompt_tokens": 128011, "total_s": 913.6971694999957, "ttft_s": 552.7236815490178, "user_number": 2, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.5864186649340913, "effective_prefill_tps": 173.26179234641745, "needle_correct": 3, "needle_total": 3, "output_tokens": 105, "prompt_tokens": 128011, "total_s": 916.1776395619963, "ttft_s": 738.8299420570256, "user_number": 1, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.174087344255268, "effective_prefill_tps": 141.73845190944695, "needle_correct": 3, "needle_total": 3, "output_tokens": 106, "prompt_tokens": 128011, "total_s": 917.7854235619889, "ttft_s": 903.1494155289838, "user_number": 0, "valid_json": null } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 128000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "2a73804cb33351a1318b0917ba061450", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 128000, "document": "distinct", "input_tokens": { "max": 128011, "median": 128011, "min": 128011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "8ed191b45f29460ab5adc9f9", "measured_at": "2026-10-06T06:48:47.036077+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 365.8734302090015 }, { "basis": null, "n": 1, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 361.40095507900696 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 1, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 361.40095507900696 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 361.40095507900696 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 365.86349211697234 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 19.71972428493775 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 691.7828978768038 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 89 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 19.71971580219204 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.24325352062094163 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 116.08256149291992 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "1/1" }, { "met": true, "name": "hidden-facts", "observed": "3/3" }, { "met": false, "name": "first-token-300-s-reference", "observed": 361.40095507900696 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 3, "skipped": null, "suite": "hidden-facts", "total": 3, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 19.71972428493775, "effective_prefill_tps": 691.7828978768038, "needle_correct": 3, "needle_total": 3, "output_tokens": 89, "prompt_tokens": 250011, "total_s": 365.86349211697234, "ttft_s": 361.40095507900696, "user_number": 0, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "10597aa97c3a6a4e114a3fa137d2bc6c", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp CUDA", "recipe": "with multi-token prediction", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "a17c165b6ba54db2932716bb", "measured_at": "2026-10-06T11:53:21.372176+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 570.2143096750369 }, { "basis": null, "n": 1, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 562.3447651329916 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 1, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 562.3447651329916 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 562.3447651329916 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 570.176155727997 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.024506792580711 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 444.58669396677624 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 13.02450476962477 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.1806338393343712 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 75.58417129516602 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "IQ3_S GGUF variant", "weights_dtype": "IQ3_S", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "1/1" }, { "met": true, "name": "hidden-facts", "observed": "3/3" }, { "met": false, "name": "first-token-300-s-reference", "observed": 562.3447651329916 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 3, "skipped": null, "suite": "hidden-facts", "total": 3, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 13.024506792580711, "effective_prefill_tps": 444.58669396677624, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 570.176155727997, "ttft_s": 562.3447651329916, "user_number": 0, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "q8_0", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "fa4cd389621fd181fb8cb62d87434f73", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp CUDA", "recipe": "with multi-token prediction", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "a4cda7bcae334b49a64c239a", "measured_at": "2026-10-06T14:08:04.866433+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 2893.4808454909944 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 1733.375218783971 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 2885.0575637820293 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 568.0434467869927 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 676.2464974020259 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9981234298489832 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 440.1265456262717 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 1148.7462295359583 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 1266.1820665659616 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9281664162886831 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 217.63814633018922 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 110 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 1733.375218783971 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 1846.7927926039556 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9434164071418897 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 144.23363002465922 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 2313.2723165829666 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 2422.1491850019665 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0195002998509235 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 108.07676995387294 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 2885.0575637820293 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 2893.4409196469933 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.166965311144942 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 86.65719642427513 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.23092824175446966 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.18731763883787803 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 85.23461532592773 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "IQ3_S GGUF variant", "weights_dtype": "IQ3_S", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 2885.0575637820293 } ], "reasons": [ "Observed decode overlap was 1 of 5 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.9981234298489832, "effective_prefill_tps": 440.1265456262717, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 676.2464974020259, "ttft_s": 568.0434467869927, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.9281664162886831, "effective_prefill_tps": 217.63814633018922, "needle_correct": 3, "needle_total": 3, "output_tokens": 110, "prompt_tokens": 250011, "total_s": 1266.1820665659616, "ttft_s": 1148.7462295359583, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.9434164071418897, "effective_prefill_tps": 144.23363002465922, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "total_s": 1846.7927926039556, "ttft_s": 1733.375218783971, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 1.0195002998509235, "effective_prefill_tps": 108.07676995387294, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "total_s": 2422.1491850019665, "ttft_s": 2313.2723165829666, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 12.166965311144942, "effective_prefill_tps": 86.65719642427513, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 2893.4409196469933, "ttft_s": 2885.0575637820293, "user_number": 3, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "q8_0", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "fa4cd389621fd181fb8cb62d87434f73", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "plain", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "b5dc5ee14f6d41568d60e290", "measured_at": "2026-10-06T10:17:56.287020+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 1477.5605192850344 }, { "basis": null, "n": 4, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 914.1351844664896 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": 4, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 1453.7408197799814 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 368.5069863999961 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 1473.6382492879638 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.07962854997879194 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 678.443039689417 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 89 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 1453.7408197799814 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 1476.3287342679687 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.51568913341892 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 171.97769822397797 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 733.6020172639983 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 1476.32891044399 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.1373317715254474 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 340.799226442189 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 1094.668351668981 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 1477.5514783319668 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.2925174608140476 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 228.38972152508285 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 113 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.3642776005404601 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.2761308214958424 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 114.21316146850586 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "4/4" }, { "met": true, "name": "hidden-facts", "observed": "12/12" }, { "met": false, "name": "first-token-300-s-reference", "observed": 1453.7408197799814 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 12, "skipped": null, "suite": "hidden-facts", "total": 12, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.07962854997879194, "effective_prefill_tps": 678.443039689417, "needle_correct": 3, "needle_total": 3, "output_tokens": 89, "prompt_tokens": 250011, "total_s": 1473.6382492879638, "ttft_s": 368.5069863999961, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 4.51568913341892, "effective_prefill_tps": 171.97769822397797, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1476.3287342679687, "ttft_s": 1453.7408197799814, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.1373317715254474, "effective_prefill_tps": 340.799226442189, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1476.32891044399, "ttft_s": 733.6020172639983, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.2925174608140476, "effective_prefill_tps": 228.38972152508285, "needle_correct": 3, "needle_total": 3, "output_tokens": 113, "prompt_tokens": 250011, "total_s": 1477.5514783319668, "ttft_s": 1094.668351668981, "user_number": 3, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 4, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "4b077fafed0bf4c9ab19a70af78278b9", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011.0, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 4, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "plain", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "b9fd006b9de145efb90aa4cd", "measured_at": "2026-10-06T10:49:38.835530+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 1834.736087628 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 1082.9542353589786 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 1821.6599746940192 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 364.5191482230439 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 1462.2789254090167 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.0929165033368863 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 685.865204115483 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 725.5963718580315 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 1822.5679372310406 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.09845287098510724 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 344.5593303613108 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 1082.9542353589786 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 1823.172631630965 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.14995574354681948 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 230.8601710367993 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 1439.2175867260084 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 1823.3605083149741 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.29155815116083383 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 173.71313573838083 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 113 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 1821.6599746940192 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 1834.726327773009 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 8.112439588858512 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 137.2435050849673 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 107 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.3666150281399662 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.29650040878811024 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 114.31551742553711 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 1821.6599746940192 } ], "reasons": [ "Observed decode overlap was 4 of 5 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.0929165033368863, "effective_prefill_tps": 685.865204115483, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1462.2789254090167, "ttft_s": 364.5191482230439, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.09845287098510724, "effective_prefill_tps": 344.5593303613108, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 1822.5679372310406, "ttft_s": 725.5963718580315, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.14995574354681948, "effective_prefill_tps": 230.8601710367993, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "total_s": 1823.172631630965, "ttft_s": 1082.9542353589786, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.29155815116083383, "effective_prefill_tps": 173.71313573838083, "needle_correct": 3, "needle_total": 3, "output_tokens": 113, "prompt_tokens": 250011, "total_s": 1823.3605083149741, "ttft_s": 1439.2175867260084, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 8.112439588858512, "effective_prefill_tps": 137.2435050849673, "needle_correct": 3, "needle_total": 3, "output_tokens": 107, "prompt_tokens": 250011, "total_s": 1834.726327773009, "ttft_s": 1821.6599746940192, "user_number": 4, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "4b077fafed0bf4c9ab19a70af78278b9", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "ba811d67c0644bd88d484e12", "measured_at": "2026-10-06T09:03:08.961159+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 18.82287276396528 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 6.833186442032456 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 7.788744076970033 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 5.511889144952875 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 11.835937755997293 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 18.18455345190772 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 116 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249957 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 6.832687321992125 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 15.319429697003216 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.607900095454491 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249984 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 6.833186442032456 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 16.40479471499566 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.805748498837925 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 114 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249984 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 6.833565931010526 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 18.2984590040287 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.20506682921896 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 118 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249984 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 7.788744076970033 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 18.815700570004992 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.70349344060487 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 249920 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 42.01841155928215 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 29.963545260728107 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 115.99076461791992 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": true, "name": "first-token-300-s-reference", "observed": 7.788744076970033 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "77fda79ee1961c35a8f8a77e7d5a01d5f9ce414975e00e640a166e71e5f9093e", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 249957, "checks_passed": true, "completed": true, "decode_tps": 18.18455345190772, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 116, "prompt_tokens": 250011, "total_s": 11.835937755997293, "ttft_s": 5.511889144952875, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": 249984, "checks_passed": true, "completed": true, "decode_tps": 12.607900095454491, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "total_s": 15.319429697003216, "ttft_s": 6.832687321992125, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": 249984, "checks_passed": true, "completed": true, "decode_tps": 11.805748498837925, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 114, "prompt_tokens": 250011, "total_s": 16.40479471499566, "ttft_s": 6.833186442032456, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": 249984, "checks_passed": true, "completed": true, "decode_tps": 10.20506682921896, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 118, "prompt_tokens": 250011, "total_s": 18.2984590040287, "ttft_s": 6.833565931010526, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": 249920, "checks_passed": true, "completed": true, "decode_tps": 9.70349344060487, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "total_s": 18.815700570004992, "ttft_s": 7.788744076970033, "user_number": 1, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "warm", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "shared-record-groups-v2", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "shared", "thinking": false }, "setup_complete": false, "setup_id": "10597aa97c3a6a4e114a3fa137d2bc6c", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": 250000, "document": "shared", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp CUDA", "recipe": "with multi-token prediction", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "bd428b7f06ed4bdf99f5be85", "measured_at": "2026-10-06T14:36:53.993491+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 513.2373797660111 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 306.9715376420063 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 508.6225764299743 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 101.71951265900861 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 201.7625528749777 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0695396678170965 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 629.2892909798165 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 204.54911168501712 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 297.7006574769621 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.1271954652746043 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 312.93707155555785 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 106 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 306.9715376420063 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 406.9156182059669 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0605930776677048 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 208.52421853732366 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 107 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 409.6804535849951 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 508.84289827302564 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.1496287768887616 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 156.24616561483043 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 115 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 508.6225764299743 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 513.199388330977 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 23.597211844415128 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 125.85166873498557 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 1.3123419049815044 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 1.06188680225994 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 78.26451110839844 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "IQ3_S GGUF variant", "weights_dtype": "IQ3_S", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 508.6225764299743 } ], "reasons": [ "Observed decode overlap was 2 of 5 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "ae06db6e13f7e0df59a2831f51932b997a654b4e3402352238c514019f5351e2", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 1.0695396678170965, "effective_prefill_tps": 629.2892909798165, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 64011, "total_s": 201.7625528749777, "ttft_s": 101.71951265900861, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 1.1271954652746043, "effective_prefill_tps": 312.93707155555785, "needle_correct": 3, "needle_total": 3, "output_tokens": 106, "prompt_tokens": 64011, "total_s": 297.7006574769621, "ttft_s": 204.54911168501712, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 1.0605930776677048, "effective_prefill_tps": 208.52421853732366, "needle_correct": 3, "needle_total": 3, "output_tokens": 107, "prompt_tokens": 64011, "total_s": 406.9156182059669, "ttft_s": 306.9715376420063, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 1.1496287768887616, "effective_prefill_tps": 156.24616561483043, "needle_correct": 3, "needle_total": 3, "output_tokens": 115, "prompt_tokens": 64011, "total_s": 508.84289827302564, "ttft_s": 409.6804535849951, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 23.597211844415128, "effective_prefill_tps": 125.85166873498557, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 64011, "total_s": 513.199388330977, "ttft_s": 508.6225764299743, "user_number": 4, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 64000, "fixture_variant": "distinct-v1", "kv_dtype": "q8_0", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "fa4cd389621fd181fb8cb62d87434f73", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 64000, "document": "distinct", "input_tokens": { "max": 64011, "median": 64011, "min": 64011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "bf96ae396a254a578155179f", "measured_at": "2026-10-06T08:54:06.974994+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 239.59412658499787 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 144.94398771203123 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 235.6320208379766 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 184.09040159301367 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 189.69232315098634 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 18.56505824362836 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 347.71503264746667 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 105 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 144.94398771203123 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 235.81806088000303 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.1224323555021356 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 441.62576875678644 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 51.194604001007974 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 236.00081456796033 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.573573797519093 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 1250.3466185369787 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 107 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 98.23997113096993 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 236.00018072797684 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.740416991948422 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 651.5779602038245 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 235.6320208379766 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 239.55685478495434 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 27.51708771861896 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 271.65662702530034 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 2.7712521947989086 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 2.1995530838400694 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 116.20093154907227 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": true, "name": "first-token-300-s-reference", "observed": 235.6320208379766 } ], "reasons": [ "Observed decode overlap was 4 of 5 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "ae06db6e13f7e0df59a2831f51932b997a654b4e3402352238c514019f5351e2", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 18.56505824362836, "effective_prefill_tps": 347.71503264746667, "needle_correct": 3, "needle_total": 3, "output_tokens": 105, "prompt_tokens": 64011, "total_s": 189.69232315098634, "ttft_s": 184.09040159301367, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 1.1224323555021356, "effective_prefill_tps": 441.62576875678644, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 64011, "total_s": 235.81806088000303, "ttft_s": 144.94398771203123, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.573573797519093, "effective_prefill_tps": 1250.3466185369787, "needle_correct": 3, "needle_total": 3, "output_tokens": 107, "prompt_tokens": 64011, "total_s": 236.00081456796033, "ttft_s": 51.194604001007974, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.740416991948422, "effective_prefill_tps": 651.5779602038245, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 64011, "total_s": 236.00018072797684, "ttft_s": 98.23997113096993, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 27.51708771861896, "effective_prefill_tps": 271.65662702530034, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 64011, "total_s": 239.55685478495434, "ttft_s": 235.6320208379766, "user_number": 3, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 64000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "10597aa97c3a6a4e114a3fa137d2bc6c", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 64000, "document": "distinct", "input_tokens": { "max": 64011, "median": 64011, "min": 64011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Startup, inference completion, concurrent decode and retrieval correctness are separate gates.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "SGLang", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "c8d4a07ac48142e6941ac2fe", "measured_at": "2026-10-06T09:10:38.226468+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/0" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "did_not_start" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "long-context-concurrency", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "42d99b231361865881080179cd5bf938", "standard": "pre-standard", "system": { "driver": null, "hardware": "unrecorded hardware", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp CUDA", "recipe": "with multi-token prediction", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "d00aaae597cd47a3a14eed43", "measured_at": "2026-10-06T12:12:34.333565+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 1152.164314196969 }, { "basis": null, "n": 2, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 857.5993945685041 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": 2, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 1143.4162936820067 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 571.7824954550015 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 685.4245042730472 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9063549744612062 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 437.2484327297416 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 104 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 1143.4162936820067 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 1152.124048580008 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.747258196883351 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 218.6526476677357 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.3687482484499347 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.187473259966871 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 78.0046157836914 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "IQ3_S GGUF variant", "weights_dtype": "IQ3_S", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "2/2" }, { "met": true, "name": "hidden-facts", "observed": "6/6" }, { "met": false, "name": "first-token-300-s-reference", "observed": 1143.4162936820067 } ], "reasons": [ "Observed decode overlap was 1 of 2 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 6, "skipped": null, "suite": "hidden-facts", "total": 6, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.9063549744612062, "effective_prefill_tps": 437.2484327297416, "needle_correct": 3, "needle_total": 3, "output_tokens": 104, "prompt_tokens": 250011, "total_s": 685.4245042730472, "ttft_s": 571.7824954550015, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 12.747258196883351, "effective_prefill_tps": 218.6526476677357, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "total_s": 1152.124048580008, "ttft_s": 1143.4162936820067, "user_number": 0, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 2, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "q8_0", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "fa4cd389621fd181fb8cb62d87434f73", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011.0, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Startup, inference completion, concurrent decode and retrieval correctness are separate gates.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "SGLang", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "d095fbb6f1b146739151bad8", "measured_at": "2026-10-06T06:29:41.821226+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/0" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "did_not_start" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "long-context-concurrency", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "42d99b231361865881080179cd5bf938", "standard": "pre-standard", "system": { "driver": null, "hardware": "unrecorded hardware", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp CUDA", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "e3ccb9cfe80f4c9886b9c79f", "measured_at": "2026-10-06T16:19:23.005448+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 597.5820526210009 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 357.17180761299096 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 592.6051474880078 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 119.95440656901337 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 472.34433928801445 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.25539889663013393 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 2084.258570827561 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 91 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[0].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 122.53851105598733 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 239.20536942401668 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 593.1022284050123 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.26843979422010505 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 1045.1939293921967 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 96 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[1].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 241.85392884601606 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 357.17180761299096 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 594.9258719370118 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.37433639779426525 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 699.9880580465682 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 90 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[2].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 359.88290120399324 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 475.3681761950138 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 596.5420531159616 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7344817403016856 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 525.9418120943673 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 90 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[3].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 478.0346490710508 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 592.6051474880078 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 597.5701633960125 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 18.32823936239311 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 421.89306161074046 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 92 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[4].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 592.70015814004 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.9505557746438538 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.7680953569251643 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 35.25088882446289 } ], "model": { "artifact_sha256": [], "publisher_repo": "nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16", "publisher_revision": "a9904d24bcc1d289a1950fa9d2b978c47cf903b9", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/5" }, { "met": false, "name": "hidden-facts", "observed": "0/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 592.6051474880078 } ], "reasons": [ "Observed decode overlap was 4 of 5 requested users.", "One or more requests did not meet completion, exact retrieval, synthesis-presence, reasoning-off or token-accounting requirements." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "04cf656671acc024178598a26462c2de38b6f97137ea5abf44a09b83b8047c5e", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 0, "checks_passed": false, "completed": true, "decode_tps": 0.25539889663013393, "effective_prefill_tps": 2084.258570827561, "first_real_token_from_dispatch_s": 122.53851105598733, "first_token_within_300_seconds": true, "needle_correct": 0, "needle_total": 3, "output_tokens": 91, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": false, "total_s": 472.34433928801445, "ttft_s": 119.95440656901337, "user_number": 3, "valid_json": false }, { "cached_prompt_tokens": 0, "checks_passed": false, "completed": true, "decode_tps": 0.26843979422010505, "effective_prefill_tps": 1045.1939293921967, "first_real_token_from_dispatch_s": 241.85392884601606, "first_token_within_300_seconds": true, "needle_correct": 0, "needle_total": 3, "output_tokens": 96, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": false, "total_s": 593.1022284050123, "ttft_s": 239.20536942401668, "user_number": 0, "valid_json": false }, { "cached_prompt_tokens": 0, "checks_passed": false, "completed": true, "decode_tps": 0.37433639779426525, "effective_prefill_tps": 699.9880580465682, "first_real_token_from_dispatch_s": 359.88290120399324, "first_token_within_300_seconds": false, "needle_correct": 0, "needle_total": 3, "output_tokens": 90, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": false, "total_s": 594.9258719370118, "ttft_s": 357.17180761299096, "user_number": 1, "valid_json": false }, { "cached_prompt_tokens": 0, "checks_passed": false, "completed": true, "decode_tps": 0.7344817403016856, "effective_prefill_tps": 525.9418120943673, "first_real_token_from_dispatch_s": 478.0346490710508, "first_token_within_300_seconds": false, "needle_correct": 0, "needle_total": 3, "output_tokens": 90, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": false, "total_s": 596.5420531159616, "ttft_s": 475.3681761950138, "user_number": 2, "valid_json": false }, { "cached_prompt_tokens": 0, "checks_passed": false, "completed": true, "decode_tps": 18.32823936239311, "effective_prefill_tps": 421.89306161074046, "first_real_token_from_dispatch_s": 592.70015814004, "first_token_within_300_seconds": false, "needle_correct": 0, "needle_total": 3, "output_tokens": 92, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": false, "total_s": 597.5701633960125, "ttft_s": 592.6051474880078, "user_number": 4, "valid_json": false } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "q8_0", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "9e5118825b2267598d5691416d3ab69e", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250016, "median": 250016, "min": 250016 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp CUDA", "recipe": "with multi-token prediction", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "ef86f90ed3e74387969779a7", "measured_at": "2026-10-06T15:25:44.979983+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 39.27173627197044 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 3.3853720990009606 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 3.390290926967282 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 3.38345537398709 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 38.15429636702174 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.9910119234916817 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 105 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 250007 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 3.3853720990009606 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 38.15018672699807 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.192883413524906 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 250007 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 3.3845772329950705 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 38.95239191601286 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.120798986084407 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 250007 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 3.389081882953178 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 38.96344179700827 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.120224798651822 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 250007 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 3.390290926967282 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 39.26064213499194 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.0108431159112548 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 250007 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 15.191450148772354 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 14.004983028788404 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 85.53128814697266 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "IQ3_S GGUF variant", "weights_dtype": "IQ3_S", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": true, "name": "first-token-300-s-reference", "observed": 3.390290926967282 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "77fda79ee1961c35a8f8a77e7d5a01d5f9ce414975e00e640a166e71e5f9093e", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 250007, "checks_passed": true, "completed": true, "decode_tps": 2.9910119234916817, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 105, "prompt_tokens": 250011, "total_s": 38.15429636702174, "ttft_s": 3.38345537398709, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": 250007, "checks_passed": true, "completed": true, "decode_tps": 3.192883413524906, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "total_s": 38.15018672699807, "ttft_s": 3.3853720990009606, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": 250007, "checks_passed": true, "completed": true, "decode_tps": 3.120798986084407, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "total_s": 38.95239191601286, "ttft_s": 3.3845772329950705, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": 250007, "checks_passed": true, "completed": true, "decode_tps": 3.120224798651822, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "total_s": 38.96344179700827, "ttft_s": 3.389081882953178, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": 250007, "checks_passed": true, "completed": true, "decode_tps": 3.0108431159112548, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 39.26064213499194, "ttft_s": 3.390290926967282, "user_number": 2, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "warm", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "shared-record-groups-v2", "kv_dtype": "q8_0", "max_output_tokens": 512, "scenario": "shared", "thinking": false }, "setup_complete": false, "setup_id": "fa4cd389621fd181fb8cb62d87434f73", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": 250000, "document": "shared", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp CUDA", "recipe": "with multi-token prediction", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "f06367aae42b4b7fa6a71d40", "measured_at": "2026-10-06T12:41:18.414586+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 1722.8193667429732 }, { "basis": null, "n": 3, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 1147.1964588109986 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": 3, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 1714.0760957379825 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 571.5261231710319 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 680.9404873010353 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9870733231303811 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 437.4445714097009 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 1147.1964588109986 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 1243.5722049530013 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.1102378376645008 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 217.93215807092156 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 1714.0760957379825 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 1722.7794267410063 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.753737616256988 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 145.8575850988457 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.2831696088176408 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.19096604458422214 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 80.5087890625 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "IQ3_S GGUF variant", "weights_dtype": "IQ3_S", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "3/3" }, { "met": true, "name": "hidden-facts", "observed": "9/9" }, { "met": false, "name": "first-token-300-s-reference", "observed": 1714.0760957379825 } ], "reasons": [ "Observed decode overlap was 1 of 3 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 9, "skipped": null, "suite": "hidden-facts", "total": 9, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.9870733231303811, "effective_prefill_tps": 437.4445714097009, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 680.9404873010353, "ttft_s": 571.5261231710319, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 1.1102378376645008, "effective_prefill_tps": 217.93215807092156, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "total_s": 1243.5722049530013, "ttft_s": 1147.1964588109986, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 12.753737616256988, "effective_prefill_tps": 145.8575850988457, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "total_s": 1722.7794267410063, "ttft_s": 1714.0760957379825, "user_number": 0, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 3, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "q8_0", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "fa4cd389621fd181fb8cb62d87434f73", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 3, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "SGLang", "recipe": "speculative decoding", "version": "0.0.0.dev0+qwen38.27b.g561c8f3" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "f08bccbb9f9c4690a60258d8", "measured_at": "2026-10-06T08:18:00.543157+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 1822.4357442200417 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 1090.2684326329618 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 1817.3838160760351 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 1090.2684326329618 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 1455.0757228239672 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.2795996756166658 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 229.31141773611824 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 103 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 368.053490760969 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 1459.6651308879955 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.0806147504892489 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 679.2789805717907 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 89 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 730.5512792259688 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 1817.6967177399783 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.0984228937631891 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 342.22238343746494 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 1445.4728020929615 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 1817.9687187419622 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.28993606418984547 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 172.96140033766008 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 1817.3838160760351 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 1822.3951647760114 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 22.349272961295103 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 137.56642806460434 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 113 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.3554872056493997 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.28642985172758634 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 116.26244354248047 } ], "model": { "artifact_sha256": [ "2228b9b22e93a88d84556419c879448ab6c490ae65c4c0b166f4962190ddbf26", "d3cfb92742e30c8b46564665791dbe0a86ed64cfc02b1275081530793c0c9581", "db6146a5464fb0a891181b93c81593f0ca65c602eb14120a1c2b1b09bca11f85", "fbcdb5ba1cdda462b5f38592d071e772c4d398afea61a0aa9188b32d1a239a79" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 1817.3838160760351 } ], "reasons": [ "Observed decode overlap was 4 of 5 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.2795996756166658, "effective_prefill_tps": 229.31141773611824, "needle_correct": 3, "needle_total": 3, "output_tokens": 103, "prompt_tokens": 250011, "total_s": 1455.0757228239672, "ttft_s": 1090.2684326329618, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.0806147504892489, "effective_prefill_tps": 679.2789805717907, "needle_correct": 3, "needle_total": 3, "output_tokens": 89, "prompt_tokens": 250011, "total_s": 1459.6651308879955, "ttft_s": 368.053490760969, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.0984228937631891, "effective_prefill_tps": 342.22238343746494, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "total_s": 1817.6967177399783, "ttft_s": 730.5512792259688, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.28993606418984547, "effective_prefill_tps": 172.96140033766008, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 1817.9687187419622, "ttft_s": 1445.4728020929615, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 22.349272961295103, "effective_prefill_tps": 137.56642806460434, "needle_correct": 3, "needle_total": 3, "output_tokens": 113, "prompt_tokens": 250011, "total_s": 1822.3951647760114, "ttft_s": 1817.3838160760351, "user_number": 1, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8_e4m3", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "10597aa97c3a6a4e114a3fa137d2bc6c", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "f494ed66edf94dd08f2efc6e", "measured_at": "2026-10-06T04:50:33.437416+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 25.812098783964757 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 8.910158696991857 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 10.684269275981933 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 4.6904168230248615 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 25.517841073975433 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.185470785955245 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 8.909450938052032 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 25.67022569302935 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.443616215767595 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 8.910158696991857 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 25.673749656998552 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.442533718321941 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 10.682348748960067 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 25.797031133959536 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.145370127471838 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 10.684269275981933 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 25.796638077998068 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.146464026578796 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 25.584141485107594 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 21.11412963980171 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 103.49654006958008 } ], "model": { "artifact_sha256": [ "1d8268aa85ace093a561e3e7b63b9d390dac1cd55a90cd55b5ec509c3c9da9fe", "f078f51aed0c1dba613c1759487004ee06aba1a290606b6be125f59ab710959e" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": true, "name": "first-token-300-s-reference", "observed": 10.684269275981933 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "2d23afb10cda3f1d9b23f9375d8c3456fc82a4da61760dec43722fa7b44f92f8", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.185470785955245, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 25.517841073975433, "ttft_s": 4.6904168230248615, "user_number": 0, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 6.443616215767595, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 25.67022569302935, "ttft_s": 8.909450938052032, "user_number": 1, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 6.442533718321941, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 25.673749656998552, "ttft_s": 8.910158696991857, "user_number": 4, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.145370127471838, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 25.797031133959536, "ttft_s": 10.682348748960067, "user_number": 3, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.146464026578796, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "total_s": 25.796638077998068, "ttft_s": 10.684269275981933, "user_number": 2, "valid_json": null } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "warm", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "shared-question-order-v1", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "shared", "thinking": false }, "setup_complete": false, "setup_id": "2a73804cb33351a1318b0917ba061450", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": 250000, "document": "shared", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "ffc9ce5ad3da4c17af151704", "measured_at": "2026-10-06T04:40:41.948441+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 391.83902825100813 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 243.79280435701367 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 379.6172779889894 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 87.2323192069889 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 387.78093562897993 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.3293976235145836 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 733.7991306652265 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 100 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 169.65645615803078 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 389.39248525799485 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.4732951643204924 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 377.2977548250527 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 105 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 243.79280435701367 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 389.9210923779756 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.6911734980807527 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 262.5631226845456 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 102 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 318.051010383002 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 390.3939307410037 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.3546591638135381 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 201.26016868463003 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 99 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 379.6172779889894 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 391.8317212719703 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 8.76012090940641 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 168.6198276830187 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 64011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 1.6710506855578764 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 1.3117631551258768 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 103.6269760131836 } ], "model": { "artifact_sha256": [ "1d8268aa85ace093a561e3e7b63b9d390dac1cd55a90cd55b5ec509c3c9da9fe", "f078f51aed0c1dba613c1759487004ee06aba1a290606b6be125f59ab710959e" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 379.6172779889894 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-06", "reproduce": { "driver": null, "fixture_sha256": "ae06db6e13f7e0df59a2831f51932b997a654b4e3402352238c514019f5351e2", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.3293976235145836, "effective_prefill_tps": 733.7991306652265, "needle_correct": 3, "needle_total": 3, "output_tokens": 100, "prompt_tokens": 64011, "total_s": 387.78093562897993, "ttft_s": 87.2323192069889, "user_number": 2, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.4732951643204924, "effective_prefill_tps": 377.2977548250527, "needle_correct": 3, "needle_total": 3, "output_tokens": 105, "prompt_tokens": 64011, "total_s": 389.39248525799485, "ttft_s": 169.65645615803078, "user_number": 0, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 0.6911734980807527, "effective_prefill_tps": 262.5631226845456, "needle_correct": 3, "needle_total": 3, "output_tokens": 102, "prompt_tokens": 64011, "total_s": 389.9210923779756, "ttft_s": 243.79280435701367, "user_number": 1, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 1.3546591638135381, "effective_prefill_tps": 201.26016868463003, "needle_correct": 3, "needle_total": 3, "output_tokens": 99, "prompt_tokens": 64011, "total_s": 390.3939307410037, "ttft_s": 318.051010383002, "user_number": 3, "valid_json": null }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 8.76012090940641, "effective_prefill_tps": 168.6198276830187, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 64011, "total_s": 391.8317212719703, "ttft_s": 379.6172779889894, "user_number": 4, "valid_json": null } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 64000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "2a73804cb33351a1318b0917ba061450", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 64000, "document": "distinct", "input_tokens": { "max": 64011, "median": 64011, "min": 64011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Finite lab cohort; configured 262144 context differs from measured inputs. Aggregate uses first-output to final-completion window. Whole-host memory sampled during speed only; missing counters remain unassessed. Broader coding quality and delivery acceptance unassessed." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "024e2d0373e84abfb3875bf4", "measured_at": "2026-10-07T05:44:16.608851+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 162.72283291816711 }, { "basis": null, "n": 16, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 93.55894066250767 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 48 }, { "basis": null, "n": 16, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 129.4477982899989 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 39 }, { "basis": null, "n": null, "name": "preemptions", "scope": "machine_total", "source_field": "preemptions", "statistic": "single", "unit": "count", "value": 0.0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "correctness_probes_passed", "scope": "machine_total", "source_field": "correctness_probes_passed", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 79.25596274786676 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 50.34327299426833 }, { "basis": null, "n": null, "name": "correctness_probes_total", "scope": "machine_total", "source_field": "correctness_probes_total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": 16, "name": "decode_tps", "scope": "per_person", "source_field": "per_request_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 8.85223287938194 }, { "basis": null, "n": null, "name": "forced_output_tokens", "scope": "machine_total", "source_field": "forced_output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 94.5468864440918 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 119.6756575719919 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[0].total_s", "statistic": "single", "unit": "s", "value": 162.35709048603894 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.972419038251711 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 62.84592934302054 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[1].total_s", "statistic": "single", "unit": "s", "value": 160.66790649900213 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.223775013105565 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 62.84495807701023 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[2].total_s", "statistic": "single", "unit": "s", "value": 160.66760278696893 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.223739365410741 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[2].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 115.75592758500716 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[3].total_s", "statistic": "single", "unit": "s", "value": 162.2341767050093 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.994390057178135 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 125.59252984198974 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[4].total_s", "statistic": "single", "unit": "s", "value": 162.55665049800882 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.824216319259065 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40107 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[5].ttft_s", "statistic": "single", "unit": "s", "value": 106.4212747340207 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[5].total_s", "statistic": "single", "unit": "s", "value": 161.97364485898288 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.198527422872003 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[5].user_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[5].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[5].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[5].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[6].ttft_s", "statistic": "single", "unit": "s", "value": 62.8436558979447 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[6].total_s", "statistic": "single", "unit": "s", "value": 160.66437300498364 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.22384230163479 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[6].user_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[6].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[6].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[6].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[7].ttft_s", "statistic": "single", "unit": "s", "value": 93.55851541401353 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[7].total_s", "statistic": "single", "unit": "s", "value": 161.61167190701235 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.508836126544265 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[7].user_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40101 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[7].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[7].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[7].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[8].ttft_s", "statistic": "single", "unit": "s", "value": 118.84896525798831 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[8].total_s", "statistic": "single", "unit": "s", "value": 162.33319285995094 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.751387300183673 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[8].user_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40106 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[8].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[8].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[8].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[9].ttft_s", "statistic": "single", "unit": "s", "value": 128.74579243297921 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[9].total_s", "statistic": "single", "unit": "s", "value": 162.67787008499727 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 15.059496363306511 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[9].user_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40101 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[9].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[9].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[9].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[10].ttft_s", "statistic": "single", "unit": "s", "value": 93.5593659110018 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[10].total_s", "statistic": "single", "unit": "s", "value": 161.60899732500548 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.509225096182421 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[10].user_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[10].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[10].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[10].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[11].ttft_s", "statistic": "single", "unit": "s", "value": 62.844671947997995 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[11].total_s", "statistic": "single", "unit": "s", "value": 160.66355964302784 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.22393999810288 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[11].user_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40106 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[11].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[11].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[11].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[12].ttft_s", "statistic": "single", "unit": "s", "value": 59.561850305995904 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[12].total_s", "statistic": "single", "unit": "s", "value": 160.53207027696772 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.060898155385902 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[12].user_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[12].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[12].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[12].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[13].ttft_s", "statistic": "single", "unit": "s", "value": 129.4477982899989 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[13].total_s", "statistic": "single", "unit": "s", "value": 162.721182977024 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 15.357620055986176 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[13].user_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[13].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[13].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[13].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[14].ttft_s", "statistic": "single", "unit": "s", "value": 92.11044140299782 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[14].total_s", "statistic": "single", "unit": "s", "value": 161.56033048901008 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.357823125781194 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[14].user_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[14].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[14].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[14].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[15].ttft_s", "statistic": "single", "unit": "s", "value": 61.292814782995265 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[15].total_s", "statistic": "single", "unit": "s", "value": 160.60114815901034 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.1455903309260105 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[15].user_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40107 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[15].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[15].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[15].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 16, "name": "decode_tps", "scope": "per_person", "source_field": "per_person_decode_median_tps", "statistic": "median", "unit": "tok/s", "value": 7.509030611363343 } ], "model": { "artifact_sha256": [ "648fe83171485f4be03ea58d86089e088c94c2a3dc88922233f9002f30d8bc8b", "8303cef7d375021d69dcafc60d3fad5f8186995b18b0db5e2a609aadeb8a4f51", "cb5c5978721a1e87406ae68f1cd5a855326e40ff8bc13832644c392115336987" ], "publisher_repo": "ornith-ai/Ornith-1.5-35B-A3B", "publisher_revision": "10fbf86fed7ecee4a061f8b499a618f46001cac1", "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "16/16" }, { "met": false, "name": "hidden-facts", "observed": "39/48" }, { "met": false, "name": "correctness-probes", "observed": "13/16" } ], "reasons": [ "39/48 Hidden facts found.", "13/16 probes passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 39, "skipped": null, "suite": "hidden-facts", "total": 48, "version": "1" }, { "passed": 13, "skipped": null, "suite": "correctness-probes", "total": 16, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "e2097959b1ca0142f2b93edc3ec411e474c0a6622800dd247af7a71bd9c39e83", "harness": { "name": "keep-team-load-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 11.972419038251711, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 162.35709048603894, "ttft_s": 119.6756575719919, "user_number": 0 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.223775013105565, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 160.66790649900213, "ttft_s": 62.84592934302054, "user_number": 1 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.223739365410741, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 160.66760278696893, "ttft_s": 62.84495807701023, "user_number": 2 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 10.994390057178135, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 162.2341767050093, "ttft_s": 115.75592758500716, "user_number": 3 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 13.824216319259065, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40107, "total_s": 162.55665049800882, "ttft_s": 125.59252984198974, "user_number": 4 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 9.198527422872003, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 161.97364485898288, "ttft_s": 106.4212747340207, "user_number": 5 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.22384230163479, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 160.66437300498364, "ttft_s": 62.8436558979447, "user_number": 6 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.508836126544265, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40101, "total_s": 161.61167190701235, "ttft_s": 93.55851541401353, "user_number": 7 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 11.751387300183673, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40106, "total_s": 162.33319285995094, "ttft_s": 118.84896525798831, "user_number": 8 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 15.059496363306511, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40101, "total_s": 162.67787008499727, "ttft_s": 128.74579243297921, "user_number": 9 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.509225096182421, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 161.60899732500548, "ttft_s": 93.5593659110018, "user_number": 10 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.22393999810288, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40106, "total_s": 160.66355964302784, "ttft_s": 62.844671947997995, "user_number": 11 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.060898155385902, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 160.53207027696772, "ttft_s": 59.561850305995904, "user_number": 12 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 15.357620055986176, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 162.721182977024, "ttft_s": 129.4477982899989, "user_number": 13 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.357823125781194, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 161.56033048901008, "ttft_s": 92.11044140299782, "user_number": 14 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.1455903309260105, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40107, "total_s": 160.60114815901034, "ttft_s": 61.292814782995265, "user_number": 15 } ], "scenario": { "id": "team-load", "version": "1" }, "settings": { "concurrency": 16, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "plain", "thinking": false }, "setup_complete": false, "setup_id": "a167e4bafd3fd96d0b47cf9d6645473d", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": { "max": 40107, "median": 40104.5, "min": 40101 }, "output_tokens": 512, "repetitions": null, "users": 16, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. No summaries. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "042d9bd08ed94402bc2085e5", "measured_at": "2026-10-07T16:31:48.247200+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 4.037177833961323 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20229 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 3.962868105038069 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.94677230783611 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20288 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 4.037177833961323 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.27615355001669 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 4.037177833961323 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 3.962868105038069 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 3.962868105038069 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 2 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "1/2" } ], "reasons": [ "1/2 requests passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "47a03fd9bbfe612c42f992e302c0e2fe", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Finite lab cohort; configured 262144 context differs from measured inputs. Aggregate uses first-output to final-completion window. Whole-host memory sampled during speed only; missing counters remain unassessed. Broader coding quality and delivery acceptance unassessed." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "071dcd0ba7b6475fa64d25d8", "measured_at": "2026-10-07T04:48:48.101985+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 191.16432785987854 }, { "basis": null, "n": 20, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 115.25076382802217 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 60 }, { "basis": null, "n": 20, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 165.80377779400442 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 60 }, { "basis": null, "n": null, "name": "preemptions", "scope": "machine_total", "source_field": "preemptions", "statistic": "single", "unit": "count", "value": 0.0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "mtp_acceptance_rate", "scope": "machine_total", "source_field": "mtp_acceptance_rate", "statistic": "single", "unit": "ratio", "value": 0.8748396554883636 }, { "basis": null, "n": null, "name": "correctness_probes_passed", "scope": "machine_total", "source_field": "correctness_probes_passed", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 76.29056289338708 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 53.56647924138762 }, { "basis": null, "n": null, "name": "correctness_probes_total", "scope": "machine_total", "source_field": "correctness_probes_total", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": 20, "name": "decode_tps", "scope": "per_person", "source_field": "per_request_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 8.862418451735458 }, { "basis": null, "n": null, "name": "forced_output_tokens", "scope": "machine_total", "source_field": "forced_output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 93.51005935668945 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 107.24706775299273 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[0].total_s", "statistic": "single", "unit": "s", "value": 189.14215367898578 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.239690626392169 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 60.74496796197491 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[1].total_s", "statistic": "single", "unit": "s", "value": 186.4632613589638 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.064643149317831 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40099 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 150.69820191897452 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[2].total_s", "statistic": "single", "unit": "s", "value": 190.6245217739488 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.798575021592837 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[2].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40099 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 59.118749120039865 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[3].total_s", "statistic": "single", "unit": "s", "value": 186.36606391402893 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.015801833046922 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 60.74396269704448 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[4].total_s", "statistic": "single", "unit": "s", "value": 186.82731717900606 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.052874402807133 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[5].ttft_s", "statistic": "single", "unit": "s", "value": 165.6927962960326 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[5].total_s", "statistic": "single", "unit": "s", "value": 191.14264387701405 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 20.078705712244346 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[5].user_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[5].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[5].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[5].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[6].ttft_s", "statistic": "single", "unit": "s", "value": 60.74574661900988 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[6].total_s", "statistic": "single", "unit": "s", "value": 186.37305180402473 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.067587052412181 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[6].user_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40100 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[6].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[6].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[6].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[7].ttft_s", "statistic": "single", "unit": "s", "value": 123.25445990305161 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[7].total_s", "statistic": "single", "unit": "s", "value": 189.4027665160247 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.7250654803577685 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[7].user_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40100 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[7].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[7].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[7].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[8].ttft_s", "statistic": "single", "unit": "s", "value": 60.745573690976016 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[8].total_s", "statistic": "single", "unit": "s", "value": 186.3696367959492 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.0676920278641315 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[8].user_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40106 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[8].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[8].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[8].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[9].ttft_s", "statistic": "single", "unit": "s", "value": 155.79878010600805 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[9].total_s", "statistic": "single", "unit": "s", "value": 190.71137753600487 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 14.63655063260776 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[9].user_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[9].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[9].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[9].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[10].ttft_s", "statistic": "single", "unit": "s", "value": 92.27036747999955 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[10].total_s", "statistic": "single", "unit": "s", "value": 188.7056777169928 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.298888952025965 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[10].user_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40099 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[10].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[10].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[10].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[11].ttft_s", "statistic": "single", "unit": "s", "value": 124.81431129196426 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[11].total_s", "statistic": "single", "unit": "s", "value": 189.47023619298125 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.9033746834849214 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[11].user_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[11].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[11].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[11].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[12].ttft_s", "statistic": "single", "unit": "s", "value": 138.59488193498692 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[12].total_s", "statistic": "single", "unit": "s", "value": 190.2552107219817 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.891535961897352 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[12].user_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[12].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[12].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[12].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[13].ttft_s", "statistic": "single", "unit": "s", "value": 90.63305058004335 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[13].total_s", "statistic": "single", "unit": "s", "value": 188.25095784501173 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.234695296355527 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[13].user_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[13].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[13].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[13].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[14].ttft_s", "statistic": "single", "unit": "s", "value": 162.5996889579692 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[14].total_s", "statistic": "single", "unit": "s", "value": 190.92960464995122 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 18.037469844805223 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[14].user_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[14].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[14].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[14].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[15].ttft_s", "statistic": "single", "unit": "s", "value": 124.8140818530228 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[15].total_s", "statistic": "single", "unit": "s", "value": 189.70104950503446 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.875233170711402 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[15].user_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[15].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[15].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[15].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[16].ttft_s", "statistic": "single", "unit": "s", "value": 165.80377779400442 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[16].total_s", "statistic": "single", "unit": "s", "value": 191.1563285649754 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 20.15576281125534 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[16].user_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40101 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[16].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[16].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[16].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[17].ttft_s", "statistic": "single", "unit": "s", "value": 57.201292753976304 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[17].total_s", "statistic": "single", "unit": "s", "value": 186.8293357240036 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.942048250455758 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[17].user_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40101 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[17].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[17].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[17].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[18].ttft_s", "statistic": "single", "unit": "s", "value": 92.27049757598434 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[18].total_s", "statistic": "single", "unit": "s", "value": 188.2484248370165 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.324140816359036 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[18].user_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40106 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[18].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[18].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[18].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[19].ttft_s", "statistic": "single", "unit": "s", "value": 147.09166702098446 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[19].total_s", "statistic": "single", "unit": "s", "value": 190.25762091501383 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.838033308715564 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[19].user_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[19].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[19].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[19].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 20, "name": "decode_tps", "scope": "per_person", "source_field": "per_person_decode_median_tps", "statistic": "median", "unit": "tok/s", "value": 6.982378053374969 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "20/20" }, { "met": true, "name": "hidden-facts", "observed": "60/60" }, { "met": true, "name": "correctness-probes", "observed": "20/20" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 60, "skipped": null, "suite": "hidden-facts", "total": 60, "version": "1" }, { "passed": 20, "skipped": null, "suite": "correctness-probes", "total": 20, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "e2097959b1ca0142f2b93edc3ec411e474c0a6622800dd247af7a71bd9c39e83", "harness": { "name": "keep-team-load-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 6.239690626392169, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 189.14215367898578, "ttft_s": 107.24706775299273, "user_number": 0 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 4.064643149317831, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40099, "total_s": 186.4632613589638, "ttft_s": 60.74496796197491, "user_number": 1 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 12.798575021592837, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40099, "total_s": 190.6245217739488, "ttft_s": 150.69820191897452, "user_number": 2 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 4.015801833046922, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 186.36606391402893, "ttft_s": 59.118749120039865, "user_number": 3 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 4.052874402807133, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 186.82731717900606, "ttft_s": 60.74396269704448, "user_number": 4 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 20.078705712244346, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 191.14264387701405, "ttft_s": 165.6927962960326, "user_number": 5 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 4.067587052412181, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40100, "total_s": 186.37305180402473, "ttft_s": 60.74574661900988, "user_number": 6 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.7250654803577685, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40100, "total_s": 189.4027665160247, "ttft_s": 123.25445990305161, "user_number": 7 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 4.0676920278641315, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40106, "total_s": 186.3696367959492, "ttft_s": 60.745573690976016, "user_number": 8 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 14.63655063260776, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 190.71137753600487, "ttft_s": 155.79878010600805, "user_number": 9 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.298888952025965, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40099, "total_s": 188.7056777169928, "ttft_s": 92.27036747999955, "user_number": 10 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.9033746834849214, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 189.47023619298125, "ttft_s": 124.81431129196426, "user_number": 11 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 9.891535961897352, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 190.2552107219817, "ttft_s": 138.59488193498692, "user_number": 12 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.234695296355527, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 188.25095784501173, "ttft_s": 90.63305058004335, "user_number": 13 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 18.037469844805223, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 190.92960464995122, "ttft_s": 162.5996889579692, "user_number": 14 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.875233170711402, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 189.70104950503446, "ttft_s": 124.8140818530228, "user_number": 15 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 20.15576281125534, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40101, "total_s": 191.1563285649754, "ttft_s": 165.80377779400442, "user_number": 16 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.942048250455758, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40101, "total_s": 186.8293357240036, "ttft_s": 57.201292753976304, "user_number": 17 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.324140816359036, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40106, "total_s": 188.2484248370165, "ttft_s": 92.27049757598434, "user_number": 18 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 11.838033308715564, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 190.25762091501383, "ttft_s": 147.09166702098446, "user_number": 19 } ], "scenario": { "id": "team-load", "version": "1" }, "settings": { "concurrency": 20, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "cf0bbc6439f3c0ed87c21751d34d5655", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": { "max": 40106, "median": 40102.5, "min": 40099 }, "output_tokens": 512, "repetitions": null, "users": 20, "warmup": null } }, { "caveats": [ "Measured retrieval only, with no customer qualification threshold. Exact dense float32 array size excludes payloads and vector database overhead. One timing run; no reranking, ingestion, answer generation or GPU measurement.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Transformers CPU", "recipe": "plain", "version": "5.19.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "0791e5c1cbba43cabcfa102e", "measured_at": "2026-10-07T14:39:45.948052+00:00", "metrics": [ { "basis": null, "n": null, "name": "passages_per_second", "scope": "machine_total", "source_field": "passages_per_second", "statistic": "single", "unit": "passages/s", "value": 12.25921414124026 }, { "basis": null, "n": null, "name": "query_seconds", "scope": "per_person", "source_field": "query_seconds", "statistic": "single", "unit": "s", "value": 2.574559000087902 }, { "basis": null, "n": null, "name": "recall_at_5", "scope": "machine_total", "source_field": "recall_at_5", "statistic": "single", "unit": "ratio", "value": 0.99 }, { "basis": null, "n": null, "name": "mrr", "scope": "machine_total", "source_field": "mrr", "statistic": "single", "unit": "ratio", "value": 0.93 }, { "basis": null, "n": null, "name": "questions", "scope": "machine_total", "source_field": "questions", "statistic": "single", "unit": "count", "value": 100 }, { "basis": null, "n": null, "name": "passage_seconds", "scope": "machine_total", "source_field": "passage_seconds", "statistic": "single", "unit": "s", "value": 32.62851887498982 }, { "basis": null, "n": null, "name": "vector_bytes", "scope": "machine_total", "source_field": "vector_bytes", "statistic": "single", "unit": "bytes", "value": 409600 }, { "basis": null, "n": null, "name": "passages", "scope": "machine_total", "source_field": "passages", "statistic": "single", "unit": "count", "value": 400 }, { "basis": null, "n": null, "name": "index_file_bytes", "scope": "machine_total", "source_field": "index_file_bytes", "statistic": "single", "unit": "bytes", "value": 409728 }, { "basis": null, "n": null, "name": "recall_at_10", "scope": "machine_total", "source_field": "recall_at_10", "statistic": "single", "unit": "ratio", "value": 1.0 }, { "basis": null, "n": null, "name": "truncated_passages", "scope": "machine_total", "source_field": "truncated_passages", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "google/embeddinggemma-2", "publisher_revision": "914f7f89142e33e77833254d9c9b90c3cef7303b", "variant": "FP32 variant", "weights_dtype": "FP32", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-embedding-eval-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "embeddings", "version": "1" }, "settings": { "context_tokens": 262144, "corpus_sha256": "d7fb630b25a7d5128c44491d1a8f21bde479711f79520f1ce95ea8bd8a0ae1f6", "dimensions": 256, "dtype": "float32" }, "setup_complete": false, "setup_id": "9590ac8ab482fc9da9f5a77a6f2c8ed2", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple Silicon laptop CPU, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Finite lab cohort; configured 262144 context differs from measured inputs. Aggregate uses first-output to final-completion window. Whole-host memory sampled during speed only; missing counters remain unassessed. Broader coding quality and delivery acceptance unassessed." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "0bf8bb92f7a54dcab6b8fbb7", "measured_at": "2026-10-07T04:12:37.552381+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 162.24340891838074 }, { "basis": null, "n": 16, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 93.09885002250667 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 48 }, { "basis": null, "n": 16, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 128.85872771800496 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 48 }, { "basis": null, "n": null, "name": "preemptions", "scope": "machine_total", "source_field": "preemptions", "statistic": "single", "unit": "count", "value": 0.0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "correctness_probes_passed", "scope": "machine_total", "source_field": "correctness_probes_passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 79.1883537268389 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 50.492035729606265 }, { "basis": null, "n": null, "name": "correctness_probes_total", "scope": "machine_total", "source_field": "correctness_probes_total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": 16, "name": "decode_tps", "scope": "per_person", "source_field": "per_request_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 8.83934808908921 }, { "basis": null, "n": null, "name": "forced_output_tokens", "scope": "machine_total", "source_field": "forced_output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 94.34637069702148 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 119.10861641797237 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[0].total_s", "statistic": "single", "unit": "s", "value": 161.87334378500236 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.949099911577173 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 105.90154287003679 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[1].total_s", "statistic": "single", "unit": "s", "value": 161.49713731801603 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.19137577489422 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40107 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 62.31990692904219 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[2].total_s", "statistic": "single", "unit": "s", "value": 160.1821003840305 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.221628311805972 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[2].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 125.01278584997635 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[3].total_s", "statistic": "single", "unit": "s", "value": 162.07575180195272 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.78734774389396 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40099 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 62.31987140898127 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[4].total_s", "statistic": "single", "unit": "s", "value": 160.18375615496188 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.22153807123406 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[5].ttft_s", "statistic": "single", "unit": "s", "value": 115.20612059801351 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[5].total_s", "statistic": "single", "unit": "s", "value": 161.75296552001964 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.978187691480086 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[5].user_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40107 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[5].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[5].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[5].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[6].ttft_s", "statistic": "single", "unit": "s", "value": 118.30824249400757 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[6].total_s", "statistic": "single", "unit": "s", "value": 161.8481544009992 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.736358150920921 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[6].user_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[6].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[6].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[6].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[7].ttft_s", "statistic": "single", "unit": "s", "value": 62.31871002103435 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[7].total_s", "statistic": "single", "unit": "s", "value": 160.18061546003446 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.221643679506319 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[7].user_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[7].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[7].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[7].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[8].ttft_s", "statistic": "single", "unit": "s", "value": 128.85872771800496 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[8].total_s", "statistic": "single", "unit": "s", "value": 162.24276190402452 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 15.306718090229928 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[8].user_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[8].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[8].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[8].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[9].ttft_s", "statistic": "single", "unit": "s", "value": 62.31772551103495 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[9].total_s", "statistic": "single", "unit": "s", "value": 160.1817466560169 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.221530793660851 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[9].user_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[9].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[9].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[9].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[10].ttft_s", "statistic": "single", "unit": "s", "value": 128.162745217036 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[10].total_s", "statistic": "single", "unit": "s", "value": 162.1968509470462 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 15.01435072376285 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[10].user_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[10].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[10].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[10].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[11].ttft_s", "statistic": "single", "unit": "s", "value": 93.09857173601631 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[11].total_s", "statistic": "single", "unit": "s", "value": 161.1347530040075 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.510709602986034 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[11].user_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[11].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[11].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[11].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[12].ttft_s", "statistic": "single", "unit": "s", "value": 91.65808796999045 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[12].total_s", "statistic": "single", "unit": "s", "value": 161.09022410598118 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.359704431376682 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[12].user_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[12].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[12].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[12].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[13].ttft_s", "statistic": "single", "unit": "s", "value": 58.99481541500427 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[13].total_s", "statistic": "single", "unit": "s", "value": 160.04902597801993 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.056691820687167 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[13].user_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[13].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[13].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[13].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[14].ttft_s", "statistic": "single", "unit": "s", "value": 93.09912830899702 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[14].total_s", "statistic": "single", "unit": "s", "value": 161.13696646899916 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.510526698368923 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[14].user_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[14].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[14].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[14].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[15].ttft_s", "statistic": "single", "unit": "s", "value": 60.74037018499803 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[15].total_s", "statistic": "single", "unit": "s", "value": 160.11499209498288 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.142157929042207 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[15].user_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[15].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[15].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[15].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 16, "name": "decode_tps", "scope": "per_person", "source_field": "per_person_decode_median_tps", "statistic": "median", "unit": "tok/s", "value": 7.5106181506774785 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "16/16" }, { "met": true, "name": "hidden-facts", "observed": "48/48" }, { "met": true, "name": "correctness-probes", "observed": "16/16" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 48, "skipped": null, "suite": "hidden-facts", "total": 48, "version": "1" }, { "passed": 16, "skipped": null, "suite": "correctness-probes", "total": 16, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "e2097959b1ca0142f2b93edc3ec411e474c0a6622800dd247af7a71bd9c39e83", "harness": { "name": "keep-team-load-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 11.949099911577173, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 161.87334378500236, "ttft_s": 119.10861641797237, "user_number": 0 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 9.19137577489422, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40107, "total_s": 161.49713731801603, "ttft_s": 105.90154287003679, "user_number": 1 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.221628311805972, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 160.1821003840305, "ttft_s": 62.31990692904219, "user_number": 2 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 13.78734774389396, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40099, "total_s": 162.07575180195272, "ttft_s": 125.01278584997635, "user_number": 3 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.22153807123406, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 160.18375615496188, "ttft_s": 62.31987140898127, "user_number": 4 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 10.978187691480086, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40107, "total_s": 161.75296552001964, "ttft_s": 115.20612059801351, "user_number": 5 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 11.736358150920921, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 161.8481544009992, "ttft_s": 118.30824249400757, "user_number": 6 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.221643679506319, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 160.18061546003446, "ttft_s": 62.31871002103435, "user_number": 7 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 15.306718090229928, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 162.24276190402452, "ttft_s": 128.85872771800496, "user_number": 8 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.221530793660851, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 160.1817466560169, "ttft_s": 62.31772551103495, "user_number": 9 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 15.01435072376285, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 162.1968509470462, "ttft_s": 128.162745217036, "user_number": 10 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.510709602986034, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 161.1347530040075, "ttft_s": 93.09857173601631, "user_number": 11 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.359704431376682, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 161.09022410598118, "ttft_s": 91.65808796999045, "user_number": 12 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.056691820687167, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 160.04902597801993, "ttft_s": 58.99481541500427, "user_number": 13 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.510526698368923, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 161.13696646899916, "ttft_s": 93.09912830899702, "user_number": 14 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.142157929042207, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 160.11499209498288, "ttft_s": 60.74037018499803, "user_number": 15 } ], "scenario": { "id": "team-load", "version": "1" }, "settings": { "concurrency": 16, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "plain", "thinking": false }, "setup_complete": false, "setup_id": "a4b96c03f67c006a12a64bbc7734ee71", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": { "max": 40107, "median": 40103.0, "min": 40099 }, "output_tokens": 512, "repetitions": null, "users": 16, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "107303a0565a43c4b3163c58", "measured_at": "2026-10-07T05:10:00.029412+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 683.6022847559652 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 676.4695380829507 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 679.3196167440037 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 653.3103375319624 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 682.547309403948 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.009887630815838 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 382.6833675163888 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 89 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[0].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 653.3024335220107 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 668.4720596069819 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 682.9644192960113 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.072165050293038 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 374.0036646363203 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 89 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[1].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 668.471511974989 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 676.4695380829507 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 683.4355538569507 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 14.068300041147799 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 369.5820519997205 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 99 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[2].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 676.457957274979 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 678.0718922290253 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 683.5493409500341 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 18.074108046011265 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 368.7086913131571 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 100 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[3].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 678.0674483600305 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 679.3196167440037 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 683.5775634370511 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 23.01578837518544 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 368.03147419517944 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 99 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[4].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 679.3174160729977 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 15.553300437149046 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.6963113064637052 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 93.87944793701172 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 679.3196167440037 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.009887630815838, "effective_prefill_tps": 382.6833675163888, "first_real_token_from_dispatch_s": 653.3024335220107, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 89, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 682.547309403948, "ttft_s": 653.3103375319624, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 6.072165050293038, "effective_prefill_tps": 374.0036646363203, "first_real_token_from_dispatch_s": 668.471511974989, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 89, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 682.9644192960113, "ttft_s": 668.4720596069819, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 14.068300041147799, "effective_prefill_tps": 369.5820519997205, "first_real_token_from_dispatch_s": 676.457957274979, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 99, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 683.4355538569507, "ttft_s": 676.4695380829507, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 18.074108046011265, "effective_prefill_tps": 368.7086913131571, "first_real_token_from_dispatch_s": 678.0674483600305, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 100, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 683.5493409500341, "ttft_s": 678.0718922290253, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 23.01578837518544, "effective_prefill_tps": 368.03147419517944, "first_real_token_from_dispatch_s": 679.3174160729977, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 99, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 683.5775634370511, "ttft_s": 679.3196167440037, "user_number": 2, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "distinct", "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "d0a27f1110ad464348786ff3b4c49fa3", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Bounded fictional tasks only. Unsupported cases are separately counted, never passes. Original publisher name with variant; exact variant repository and revision retained privately. Not customer acceptance.", "Publisher revision was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "18057fc2b4e64d4f8cca1ee2", "measured_at": "2026-10-07T06:50:22.362351+00:00", "metrics": [ { "basis": null, "n": null, "name": "skipped_cases", "scope": "machine_total", "source_field": "skipped_cases", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "correctness_failures", "scope": "machine_total", "source_field": "correctness_failures", "statistic": "single", "unit": "count", "value": 48 }, { "basis": null, "n": null, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 0.06402582296868786 }, { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 130 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 73 }, { "basis": null, "n": null, "name": "api_cases_passed", "scope": "machine_total", "source_field": "api_cases_passed", "statistic": "single", "unit": "count", "value": 72 }, { "basis": null, "n": null, "name": "ttft_p95_s", "scope": "per_person", "source_field": "ttft_p95_s", "statistic": "p95", "unit": "s", "value": 0.1146105490042828 }, { "basis": null, "n": null, "name": "completion_p95_s", "scope": "per_person", "source_field": "completion_p95_s", "statistic": "p95", "unit": "s", "value": 0.6232617410132661 }, { "basis": null, "n": null, "name": "api_cases_total", "scope": "machine_total", "source_field": "api_cases_total", "statistic": "single", "unit": "count", "value": 124 }, { "basis": null, "n": null, "name": "completion_median_s", "scope": "per_person", "source_field": "completion_median_s", "statistic": "median", "unit": "s", "value": 0.19108319201041013 }, { "basis": null, "n": null, "name": "completeness_failures", "scope": "machine_total", "source_field": "completeness_failures", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[0].passed", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[0].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[0].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[1].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[1].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[1].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[2].passed", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[2].total", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[2].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[3].passed", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[3].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[3].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[4].passed", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[4].total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[4].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[5].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[5].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[5].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[6].passed", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[6].total", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[6].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[7].passed", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[7].total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[7].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[8].passed", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[8].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[8].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[9].passed", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[9].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[9].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[10].passed", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[10].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[10].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[11].passed", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[11].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[11].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[12].passed", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[12].total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[12].skipped", "statistic": "single", "unit": "count", "value": 4 } ], "model": { "artifact_sha256": [], "publisher_repo": "ornith-ai/Ornith-1.5-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "structured-tasks", "observed": "73/130" }, { "met": false, "name": "API-cases", "observed": "72/124" } ], "reasons": [ "73/130 tasks passed.", "72/124 API cases passed." ], "status": "partial" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 73, "skipped": null, "suite": "structured-tasks", "total": 130, "version": "core-v1" }, { "passed": 72, "skipped": 4, "suite": "API-cases", "total": 124, "version": "core-v1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "42c4dc34a066f791a62c9b1dbeb97943988c746309a43a464f0417a02ae9f1da", "harness": { "name": "keep-api-suite-run-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": "core-v1", "workload_version": "1" }, "requests": [], "scenario": { "id": "api-suite", "version": "1" }, "settings": { "concurrency": 1, "context_tokens": 262144, "kv_dtype": "fp8", "speculative_decoding": "with multi-token prediction" }, "setup_complete": false, "setup_id": "d1b964b67aa6c3ae4e36cb20bed67f31", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Operator-supplied runtime manifest; the endpoint runner cannot attest hardware or model pins. Bounded task correctness is not customer acceptance.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "1abbd6f0eb7e40ea93f8f17c", "measured_at": "2026-10-07T04:49:20.243055+00:00", "metrics": [ { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 25 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[0].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[0].passed", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[0].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[1].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[1].passed", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[1].total", "statistic": "single", "unit": "count", "value": 16 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "structured-tasks", "observed": "25/32" } ], "reasons": [ "25/32 tasks passed." ], "status": "partial" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 25, "skipped": null, "suite": "structured-tasks", "total": 32, "version": null } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": { "context_tokens": 262144, "kv_dtype": "fp8", "speculative_decoding": "with multi-token prediction" }, "setup_complete": false, "setup_id": "282291e356a060d972cd0b5ba5eb4856", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "1b41e8c80c3d4cb3802c8643", "measured_at": "2026-10-07T02:06:52.136077+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 643.4567561339936 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 632.519667607965 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 635.1736541999853 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 632.5196552400012 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 642.679639309994 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.842535117288918 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 395.2620253439186 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 101 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[0].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 632.4996626980137 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 616.0835204110481 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 643.1505341390148 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.137877976700371 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 405.80699161242586 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 113 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[1].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 616.0825532760355 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 632.519667607965 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 643.1509613379603 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.064626443168223 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 395.26201761516853 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[2].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 632.5146672279807 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 633.9609259749996 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 643.207107389986 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.572344862990803 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 394.3634217132481 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[3].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 633.9516310869949 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 635.1736541999853 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 643.4447859599604 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.782878003667918 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 393.61046911634605 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 115 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[4].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 635.1607155720121 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 19.735509353127895 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.846987765385292 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 100.59363555908203 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "4/5" }, { "met": false, "name": "hidden-facts", "observed": "14/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 635.1736541999853 } ], "reasons": [ "One or more requests did not meet completion, exact retrieval, synthesis-presence, reasoning-off or token-accounting requirements." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 14, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "77fda79ee1961c35a8f8a77e7d5a01d5f9ce414975e00e640a166e71e5f9093e", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 9.842535117288918, "effective_prefill_tps": 395.2620253439186, "first_real_token_from_dispatch_s": 632.4996626980137, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 101, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 642.679639309994, "ttft_s": 632.5196552400012, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 4.137877976700371, "effective_prefill_tps": 405.80699161242586, "first_real_token_from_dispatch_s": 616.0825532760355, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 113, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 643.1505341390148, "ttft_s": 616.0835204110481, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 10.064626443168223, "effective_prefill_tps": 395.26201761516853, "first_real_token_from_dispatch_s": 632.5146672279807, "first_token_within_300_seconds": false, "needle_correct": 2, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 643.1509613379603, "ttft_s": 632.519667607965, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 11.572344862990803, "effective_prefill_tps": 394.3634217132481, "first_real_token_from_dispatch_s": 633.9516310869949, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 643.207107389986, "ttft_s": 633.9609259749996, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 13.782878003667918, "effective_prefill_tps": 393.61046911634605, "first_real_token_from_dispatch_s": 635.1607155720121, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 115, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 643.4447859599604, "ttft_s": 635.1736541999853, "user_number": 2, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "shared-record-groups-v2", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "shared", "thinking": false }, "setup_complete": false, "setup_id": "1455fb8523cb693bc00de6e3284af4dd", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "shared", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "24e87573cc634a0b9c208722", "measured_at": "2026-10-07T05:24:24.770587+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 682.2305231079808 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 676.9396951299859 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 676.9402521060547 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 676.9396951299859 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 681.9826804589829 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 21.217590974289198 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 369.32536501939495 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[0].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 676.9267538309796 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 676.9402521060547 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 681.9728933740407 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 21.65860712015733 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 369.3250611441427 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 110 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[1].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 676.9345530970022 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 672.6820814830135 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 681.9782756119967 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.510086656473856 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 371.6629398672533 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[2].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 672.6786944590276 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 676.9392962019774 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 682.101213536982 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 22.084817055657016 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 369.32558266702335 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 115 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[3].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 676.9313521999866 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 676.9402119460283 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 682.2177754470031 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 22.16932112297455 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 369.3250830546511 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 118 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[4].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 676.9392923940322 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 58.09682858035302 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.8193711378573479 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 93.94606399536133 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "3/5" }, { "met": false, "name": "hidden-facts", "observed": "13/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 676.9402521060547 } ], "reasons": [ "One or more requests did not meet completion, exact retrieval, synthesis-presence, reasoning-off or token-accounting requirements." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 13, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "77fda79ee1961c35a8f8a77e7d5a01d5f9ce414975e00e640a166e71e5f9093e", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 21.217590974289198, "effective_prefill_tps": 369.32536501939495, "first_real_token_from_dispatch_s": 676.9267538309796, "first_token_within_300_seconds": false, "needle_correct": 2, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 681.9826804589829, "ttft_s": 676.9396951299859, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 21.65860712015733, "effective_prefill_tps": 369.3250611441427, "first_real_token_from_dispatch_s": 676.9345530970022, "first_token_within_300_seconds": false, "needle_correct": 2, "needle_total": 3, "output_tokens": 110, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 681.9728933740407, "ttft_s": 676.9402521060547, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 11.510086656473856, "effective_prefill_tps": 371.6629398672533, "first_real_token_from_dispatch_s": 672.6786944590276, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 681.9782756119967, "ttft_s": 672.6820814830135, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 22.084817055657016, "effective_prefill_tps": 369.32558266702335, "first_real_token_from_dispatch_s": 676.9313521999866, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 115, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 682.101213536982, "ttft_s": 676.9392962019774, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 22.16932112297455, "effective_prefill_tps": 369.3250830546511, "first_real_token_from_dispatch_s": 676.9392923940322, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 118, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 682.2177754470031, "ttft_s": 676.9402119460283, "user_number": 0, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "shared-record-groups-v2", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "shared", "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "d0a27f1110ad464348786ff3b4c49fa3", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "shared", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Operator-supplied runtime manifest; the endpoint runner cannot attest hardware or model pins. Bounded task correctness is not customer acceptance.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "271036436cc1413aa9607744", "measured_at": "2026-10-07T05:51:31.506535+00:00", "metrics": [ { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[0].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[0].passed", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[0].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[1].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[1].passed", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[1].total", "statistic": "single", "unit": "count", "value": 16 } ], "model": { "artifact_sha256": [], "publisher_repo": "ornith-ai/Ornith-1.5-35B-A3B", "publisher_revision": "10fbf86fed7ecee4a061f8b499a618f46001cac1", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "structured-tasks", "observed": "10/32" } ], "reasons": [ "10/32 tasks passed." ], "status": "partial" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 10, "skipped": null, "suite": "structured-tasks", "total": 32, "version": null } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": { "context_tokens": 262144, "kv_dtype": "fp8", "speculative_decoding": "plain" }, "setup_complete": false, "setup_id": "83715be47e48a10a7490a3844dd70180", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "272292f28e714c4db98cd800", "measured_at": "2026-10-07T19:46:35.303920+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 29.235718090028968 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 105 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20208 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 3.9575297120027244 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.75146692190684 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40333 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 5.716695006994996 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 68.04202756561375 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.46770634468053457 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60459 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 6.984351738996338 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 64.12189880065614 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6586943217717792 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40509 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 9.564765662013087 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.81958859999683 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[3].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60637 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[4].ttft_s", "statistic": "single", "unit": "s", "value": 7.028944340010639 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 64.43178693293225 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[4].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[4].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6567607236505764 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40502 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[5].ttft_s", "statistic": "single", "unit": "s", "value": 9.599285634001717 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 68.05532020889034 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[5].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[5].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60630 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[6].ttft_s", "statistic": "single", "unit": "s", "value": 7.006433008995373 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 64.0692595437555 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140015 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[6].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[6].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6568365495629227 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40507 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[7].ttft_s", "statistic": "single", "unit": "s", "value": 12.692896176013164 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 53.84427135862067 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160018 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[7].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[7].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60636 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[8].ttft_s", "statistic": "single", "unit": "s", "value": 11.3238241559593 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 47.36840093861204 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[8].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[8].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5812189276447403 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40513 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[9].ttft_s", "statistic": "single", "unit": "s", "value": 15.255557763040997 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 49.69953628345472 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[9].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[9].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.39438293489671017 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60634 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[10].ttft_s", "statistic": "single", "unit": "s", "value": 10.7441543880268 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.37835688459876 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[10].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[10].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3937317712195363 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40511 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[11].ttft_s", "statistic": "single", "unit": "s", "value": 15.727335946983658 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 50.61613259682372 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[11].dropped_messages", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[11].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3937317712195363 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60635 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[12].ttft_s", "statistic": "single", "unit": "s", "value": 10.882967325043865 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.135438725003755 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[12].dropped_messages", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[12].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3937473428184416 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40509 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[13].ttft_s", "statistic": "single", "unit": "s", "value": 15.784446775971446 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 49.13417891307616 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[13].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[13].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.39374344980324694 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60631 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[14].ttft_s", "statistic": "single", "unit": "s", "value": 11.514838334987871 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.087560978075416 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[14].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[14].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.32816929263629774 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40507 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[15].ttft_s", "statistic": "single", "unit": "s", "value": 20.700667419994716 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 41.80348003232164 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[15].dropped_messages", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[15].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4147942920551005 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60630 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[16].ttft_s", "statistic": "single", "unit": "s", "value": 15.948973254009616 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.098274344388592 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[16].dropped_messages", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[16].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49285298813162876 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40506 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[17].ttft_s", "statistic": "single", "unit": "s", "value": 22.930588026996702 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 40.84772287502217 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[17].dropped_messages", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[17].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.2811496184176156 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60631 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[18].ttft_s", "statistic": "single", "unit": "s", "value": 16.849213333975058 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.825760561787803 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[18].dropped_messages", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[18].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49231986450819315 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40505 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[19].ttft_s", "statistic": "single", "unit": "s", "value": 22.988517620950006 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 34.60310107915711 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[19].dropped_messages", "statistic": "single", "unit": "count", "value": 36 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[19].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.2811258021022314 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[20].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[20].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60624 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[20].ttft_s", "statistic": "single", "unit": "s", "value": 16.5535195809789 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[20].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.730850320145656 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[20].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[20].dropped_messages", "statistic": "single", "unit": "count", "value": 36 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[20].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49235029764296445 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[21].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[21].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20215 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[21].ttft_s", "statistic": "single", "unit": "s", "value": 7.8101454300340265 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[21].decode_tps", "statistic": "single", "unit": "tok/s", "value": 15.546864701387117 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[21].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[21].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[21].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[22].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[22].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40338 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[22].ttft_s", "statistic": "single", "unit": "s", "value": 11.432279617991298 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[22].decode_tps", "statistic": "single", "unit": "tok/s", "value": 44.3629379107913 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[22].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40005 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[22].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[22].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5812189276447403 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[23].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[23].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60465 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[23].ttft_s", "statistic": "single", "unit": "s", "value": 9.741424662002828 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[23].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.475686560162728 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[23].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[23].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[23].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.39438293489671017 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[24].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[24].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40511 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[24].ttft_s", "statistic": "single", "unit": "s", "value": 15.697511225007474 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[24].decode_tps", "statistic": "single", "unit": "tok/s", "value": 48.87909215299378 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[24].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[24].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[24].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3937317712195363 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[25].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[25].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60634 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[25].ttft_s", "statistic": "single", "unit": "s", "value": 10.775049289979506 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[25].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.628660778927497 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[25].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[25].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[25].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3937317712195363 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[26].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[26].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40506 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[26].ttft_s", "statistic": "single", "unit": "s", "value": 15.832782799960114 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[26].decode_tps", "statistic": "single", "unit": "tok/s", "value": 49.90595749363152 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[26].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[26].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[26].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3937473428184416 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[27].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[27].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60633 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[27].ttft_s", "statistic": "single", "unit": "s", "value": 10.770137479994446 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[27].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.970706951073282 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[27].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140015 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[27].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[27].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.39374344980324694 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[28].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[28].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40508 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[28].ttft_s", "statistic": "single", "unit": "s", "value": 18.81850099604344 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[28].decode_tps", "statistic": "single", "unit": "tok/s", "value": 43.37200772529074 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[28].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160019 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[28].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[28].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.32816929263629774 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[29].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[29].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60639 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[29].ttft_s", "statistic": "single", "unit": "s", "value": 15.851792955014389 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[29].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.345060262798667 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[29].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[29].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[29].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4147942920551005 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[30].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[30].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40509 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[30].ttft_s", "statistic": "single", "unit": "s", "value": 21.282559739018325 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[30].decode_tps", "statistic": "single", "unit": "tok/s", "value": 42.27425413947526 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[30].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[30].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[30].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49285298813162876 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[31].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[31].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60637 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[31].ttft_s", "statistic": "single", "unit": "s", "value": 14.361403310962487 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[31].decode_tps", "statistic": "single", "unit": "tok/s", "value": 8.204166613153731 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[31].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[31].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[31].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.2811496184176156 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[32].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[32].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40516 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[32].ttft_s", "statistic": "single", "unit": "s", "value": 21.671179135038983 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[32].decode_tps", "statistic": "single", "unit": "tok/s", "value": 43.15592476956295 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[32].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[32].dropped_messages", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[32].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49231986450819315 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[33].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[33].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60643 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[33].ttft_s", "statistic": "single", "unit": "s", "value": 14.09879862802336 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[33].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.4668077341273795 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[33].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[33].dropped_messages", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[33].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.2811258021022314 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[34].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[34].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40506 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[34].ttft_s", "statistic": "single", "unit": "s", "value": 21.76999982498819 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[34].decode_tps", "statistic": "single", "unit": "tok/s", "value": 42.03511929545921 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[34].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[34].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[34].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49235029764296445 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[35].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[35].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60632 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[35].ttft_s", "statistic": "single", "unit": "s", "value": 10.913511283986736 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[35].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.442870977578751 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[35].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[35].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[35].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.39375123591061895 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[36].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[36].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40510 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[36].ttft_s", "statistic": "single", "unit": "s", "value": 18.7667135510128 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[36].decode_tps", "statistic": "single", "unit": "tok/s", "value": 42.94576792650669 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[36].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[36].dropped_messages", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[36].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.32814495595784476 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[37].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[37].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60639 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[37].ttft_s", "statistic": "single", "unit": "s", "value": 15.645240779034793 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[37].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.975638360120717 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[37].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[37].dropped_messages", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[37].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.41478842878244954 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[38].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[38].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40509 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[38].ttft_s", "statistic": "single", "unit": "s", "value": 21.94474613200873 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[38].decode_tps", "statistic": "single", "unit": "tok/s", "value": 42.16675862682201 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[38].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360027 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[38].dropped_messages", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[38].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49282859158241243 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[39].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[39].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60641 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[39].ttft_s", "statistic": "single", "unit": "s", "value": 17.632886138977483 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[39].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.759934096543274 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[39].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[39].dropped_messages", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[39].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.24604130755781267 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[40].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[40].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40513 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[40].ttft_s", "statistic": "single", "unit": "s", "value": 25.985662065970246 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[40].decode_tps", "statistic": "single", "unit": "tok/s", "value": 39.6132896554064 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[40].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[40].dropped_messages", "statistic": "single", "unit": "count", "value": 36 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[40].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4874129195693477 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[41].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[41].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60643 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[41].ttft_s", "statistic": "single", "unit": "s", "value": 19.274683998024557 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[41].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.924233410671389 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[41].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420032 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[41].dropped_messages", "statistic": "single", "unit": "count", "value": 36 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[41].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.394049285845038 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[42].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[42].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20213 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[42].ttft_s", "statistic": "single", "unit": "s", "value": 10.141697470040526 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[42].decode_tps", "statistic": "single", "unit": "tok/s", "value": 36.397598071587446 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[42].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[42].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[42].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[43].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[43].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40341 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[43].ttft_s", "statistic": "single", "unit": "s", "value": 15.87518325203564 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[43].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.113118233847437 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[43].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[43].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[43].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4147942920551005 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[44].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[44].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60467 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[44].ttft_s", "statistic": "single", "unit": "s", "value": 14.947069421992637 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[44].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.526687298315467 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[44].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[44].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[44].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49285298813162876 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[45].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[45].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40504 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[45].ttft_s", "statistic": "single", "unit": "s", "value": 22.338037642010022 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[45].decode_tps", "statistic": "single", "unit": "tok/s", "value": 27.91905152620158 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[45].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[45].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[45].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.2811496184176156 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[46].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[46].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60634 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[46].ttft_s", "statistic": "single", "unit": "s", "value": 16.57099251699401 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[46].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.367412527855727 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[46].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[46].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[46].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49231986450819315 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[47].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[47].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40511 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[47].ttft_s", "statistic": "single", "unit": "s", "value": 23.30581573699601 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[47].decode_tps", "statistic": "single", "unit": "tok/s", "value": 42.709265485118074 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[47].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120017 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[47].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[47].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.2811258021022314 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[48].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[48].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60641 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[48].ttft_s", "statistic": "single", "unit": "s", "value": 16.774452330020722 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[48].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.634736823182982 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[48].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140019 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[48].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[48].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49235029764296445 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[49].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[49].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40508 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[49].ttft_s", "statistic": "single", "unit": "s", "value": 15.868723998952191 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[49].decode_tps", "statistic": "single", "unit": "tok/s", "value": 48.9291975783834 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[49].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160022 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[49].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[49].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.39375123591061895 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[50].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[50].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[50].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60637 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[50].ttft_s", "statistic": "single", "unit": "s", "value": 11.400980226986576 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[50].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.322800697888207 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[50].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[50].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[50].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[50].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.32814495595784476 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[51].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[51].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[51].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40508 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[51].ttft_s", "statistic": "single", "unit": "s", "value": 20.466051666007843 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[51].decode_tps", "statistic": "single", "unit": "tok/s", "value": 41.49908439988412 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[51].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[51].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[51].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[51].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.41478842878244954 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[52].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[52].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[52].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60633 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[52].ttft_s", "statistic": "single", "unit": "s", "value": 16.38111391896382 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[52].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.391201514494213 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[52].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[52].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[52].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[52].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49282859158241243 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[53].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[53].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[53].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40503 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[53].ttft_s", "statistic": "single", "unit": "s", "value": 26.69427929201629 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[53].decode_tps", "statistic": "single", "unit": "tok/s", "value": 37.33391221409797 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[53].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[53].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[53].dropped_messages", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[53].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.24604130755781267 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[54].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[54].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[54].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60630 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[54].ttft_s", "statistic": "single", "unit": "s", "value": 20.323367278964724 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[54].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.066251392721956 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[54].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[54].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260027 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[54].dropped_messages", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[54].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49234725416017605 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[55].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[55].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[55].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40509 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[55].ttft_s", "statistic": "single", "unit": "s", "value": 29.16802976699546 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[55].decode_tps", "statistic": "single", "unit": "tok/s", "value": 36.98619319410352 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[55].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[55].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[55].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[55].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.394049285845038 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[56].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[56].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[56].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60636 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[56].ttft_s", "statistic": "single", "unit": "s", "value": 16.698239288001787 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[56].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.481701743165717 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[56].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[56].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[56].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[56].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49230769230769234 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[57].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[57].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[57].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40506 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[57].ttft_s", "statistic": "single", "unit": "s", "value": 22.981581624015234 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[57].decode_tps", "statistic": "single", "unit": "tok/s", "value": 35.10860078250099 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[57].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[57].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320030 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[57].dropped_messages", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[57].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.2811238175914161 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[58].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[58].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[58].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60631 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[58].ttft_s", "statistic": "single", "unit": "s", "value": 16.696113876998425 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[58].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.036328198052006 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[58].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[58].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340030 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[58].dropped_messages", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[58].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4923046493516126 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[59].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[59].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[59].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40508 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[59].ttft_s", "statistic": "single", "unit": "s", "value": 23.21832900197478 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[59].decode_tps", "statistic": "single", "unit": "tok/s", "value": 41.770086656659245 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[59].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[59].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360030 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[59].dropped_messages", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[59].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.28110000564684623 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[60].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[60].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[60].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60636 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[60].ttft_s", "statistic": "single", "unit": "s", "value": 16.708718668029178 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[60].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.914169842566224 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[60].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[60].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380031 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[60].dropped_messages", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[60].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4922772644395686 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[61].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[61].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[61].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40506 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[61].ttft_s", "statistic": "single", "unit": "s", "value": 23.04407823603833 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[61].decode_tps", "statistic": "single", "unit": "tok/s", "value": 40.83694376542976 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[61].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[61].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400031 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[61].dropped_messages", "statistic": "single", "unit": "count", "value": 36 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[61].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.2811317558027899 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[62].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[62].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[62].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60632 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[62].ttft_s", "statistic": "single", "unit": "s", "value": 16.715201536018867 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[62].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.51823494022069 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[62].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[62].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420031 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[62].dropped_messages", "statistic": "single", "unit": "count", "value": 36 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[62].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49230160643314974 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[63].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[63].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[63].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20214 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[63].ttft_s", "statistic": "single", "unit": "s", "value": 10.404825248988345 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[63].decode_tps", "statistic": "single", "unit": "tok/s", "value": 36.15381217748142 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[63].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[63].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[63].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[63].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[64].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[64].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[64].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40342 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[64].ttft_s", "statistic": "single", "unit": "s", "value": 15.727747226017527 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[64].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.563622519393371 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[64].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[64].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[64].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[64].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.41478842878244954 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[65].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[65].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[65].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60472 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[65].ttft_s", "statistic": "single", "unit": "s", "value": 15.671572171966545 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[65].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.153443291987305 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[65].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[65].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60005 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[65].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[65].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49282859158241243 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[66].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[66].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[66].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40502 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[66].ttft_s", "statistic": "single", "unit": "s", "value": 26.381552103965078 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[66].decode_tps", "statistic": "single", "unit": "tok/s", "value": 33.004461697196035 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[66].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[66].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80005 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[66].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[66].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.24604130755781267 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[67].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[67].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[67].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60629 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[67].ttft_s", "statistic": "single", "unit": "s", "value": 20.293243667983916 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[67].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.761320032962875 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[67].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[67].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[67].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[67].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49234725416017605 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[68].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[68].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[68].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40508 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[68].ttft_s", "statistic": "single", "unit": "s", "value": 29.235718090028968 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[68].decode_tps", "statistic": "single", "unit": "tok/s", "value": 37.283716264826694 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[68].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[68].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[68].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[68].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.394049285845038 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[69].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[69].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[69].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60634 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[69].ttft_s", "statistic": "single", "unit": "s", "value": 16.999177466961555 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[69].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.73467223969065 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[69].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[69].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[69].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[69].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49230769230769234 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[70].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[70].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[70].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40513 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[70].ttft_s", "statistic": "single", "unit": "s", "value": 23.042161166027654 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[70].decode_tps", "statistic": "single", "unit": "tok/s", "value": 40.77675820275594 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[70].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[70].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[70].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[70].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.2811238175914161 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[71].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[71].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[71].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60640 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[71].ttft_s", "statistic": "single", "unit": "s", "value": 16.993469974026084 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[71].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.938755528502059 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[71].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[71].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180016 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[71].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[71].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4923046493516126 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[72].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[72].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[72].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40519 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[72].ttft_s", "statistic": "single", "unit": "s", "value": 22.890427438018378 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[72].decode_tps", "statistic": "single", "unit": "tok/s", "value": 36.70042630993243 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[72].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[72].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200018 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[72].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[72].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.28110000564684623 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[73].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[73].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[73].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60648 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[73].ttft_s", "statistic": "single", "unit": "s", "value": 17.0235737170442 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[73].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.272964324536494 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[73].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[73].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220018 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[73].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[73].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4922772644395686 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[74].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[74].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[74].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40510 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[74].ttft_s", "statistic": "single", "unit": "s", "value": 22.884295522992034 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[74].decode_tps", "statistic": "single", "unit": "tok/s", "value": 36.8199704699588 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[74].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[74].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240018 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[74].dropped_messages", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[74].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.2811317558027899 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[75].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[75].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[75].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60643 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[75].ttft_s", "statistic": "single", "unit": "s", "value": 17.018279869982507 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[75].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.668052638930005 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[75].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[75].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[75].dropped_messages", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[75].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49230160643314974 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[76].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[76].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[76].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40512 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[76].ttft_s", "statistic": "single", "unit": "s", "value": 15.63626259699231 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[76].decode_tps", "statistic": "single", "unit": "tok/s", "value": 49.26023502851333 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[76].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[76].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[76].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[76].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3937084161303398 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[77].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[77].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[77].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60642 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[77].ttft_s", "statistic": "single", "unit": "s", "value": 10.807475946028717 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[77].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.534510636198561 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[77].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[77].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[77].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[77].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3937084161303398 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[78].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[78].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[78].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40513 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[78].ttft_s", "statistic": "single", "unit": "s", "value": 15.772639085014816 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[78].decode_tps", "statistic": "single", "unit": "tok/s", "value": 49.03793333659132 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[78].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[78].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[78].dropped_messages", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[78].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3937200933285878 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[79].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[79].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[79].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60642 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[79].ttft_s", "statistic": "single", "unit": "s", "value": 10.910131361044478 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[79].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.30127918806825 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[79].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[79].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[79].dropped_messages", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[79].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3937006317163109 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[80].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[80].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[80].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40513 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[80].ttft_s", "statistic": "single", "unit": "s", "value": 15.840449185983744 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[80].decode_tps", "statistic": "single", "unit": "tok/s", "value": 49.209259469803456 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[80].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[80].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[80].dropped_messages", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[80].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3936928476101033 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[81].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[81].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[81].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60642 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[81].ttft_s", "statistic": "single", "unit": "s", "value": 10.933627520978916 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[81].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.212334375982705 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[81].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[81].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[81].dropped_messages", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[81].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3936694971382252 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[82].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[82].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[82].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40510 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[82].ttft_s", "statistic": "single", "unit": "s", "value": 15.847493106964976 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[82].decode_tps", "statistic": "single", "unit": "tok/s", "value": 49.40719153311712 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[82].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[82].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400026 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[82].dropped_messages", "statistic": "single", "unit": "count", "value": 36 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[82].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.39368117202791675 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[83].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[83].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[83].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60644 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[83].ttft_s", "statistic": "single", "unit": "s", "value": 10.921712554001715 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[83].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.65374719842379 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[83].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[83].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420027 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[83].dropped_messages", "statistic": "single", "unit": "count", "value": 36 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[83].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3936772803210787 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[84].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[84].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[84].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20213 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[84].ttft_s", "statistic": "single", "unit": "s", "value": 17.15052171901334 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[84].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.96666202405861 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[84].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[84].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[84].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[84].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4925421129443194 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[85].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[85].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[85].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40340 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[85].ttft_s", "statistic": "single", "unit": "s", "value": 18.720171983004548 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[85].decode_tps", "statistic": "single", "unit": "tok/s", "value": 8.601178116487572 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[85].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[85].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[85].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[85].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4874129195693477 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[86].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[86].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[86].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60467 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[86].ttft_s", "statistic": "single", "unit": "s", "value": 18.58316817699233 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[86].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.481965324238215 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[86].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[86].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[86].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[86].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.394049285845038 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[87].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[87].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[87].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40515 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[87].ttft_s", "statistic": "single", "unit": "s", "value": 21.65844379598275 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[87].decode_tps", "statistic": "single", "unit": "tok/s", "value": 41.31927637824984 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[87].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[87].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[87].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[87].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49230769230769234 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[88].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[88].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[88].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60641 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[88].ttft_s", "statistic": "single", "unit": "s", "value": 14.032035385956988 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[88].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.787351824339719 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[88].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[88].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[88].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[88].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.2811238175914161 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[89].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[89].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[89].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40515 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[89].ttft_s", "statistic": "single", "unit": "s", "value": 21.592884418962058 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[89].decode_tps", "statistic": "single", "unit": "tok/s", "value": 41.65488151674655 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[89].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[89].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[89].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[89].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4923046493516126 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[90].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[90].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[90].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60645 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[90].ttft_s", "statistic": "single", "unit": "s", "value": 14.070296765014064 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[90].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.566891605557954 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[90].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[90].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[90].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[90].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.28110000564684623 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[91].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[91].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[91].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40511 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[91].ttft_s", "statistic": "single", "unit": "s", "value": 21.671029557008296 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[91].decode_tps", "statistic": "single", "unit": "tok/s", "value": 42.153357812729894 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[91].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[91].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160017 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[91].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[91].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4922772644395686 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[92].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[92].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[92].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60640 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[92].ttft_s", "statistic": "single", "unit": "s", "value": 14.030632857989985 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[92].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.726365277915541 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[92].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[92].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[92].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[92].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.2811317558027899 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[93].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[93].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[93].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40512 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[93].ttft_s", "statistic": "single", "unit": "s", "value": 21.638244080997538 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[93].decode_tps", "statistic": "single", "unit": "tok/s", "value": 41.637943468775376 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[93].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[93].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[93].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[93].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49230160643314974 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[94].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[94].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[94].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60639 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[94].ttft_s", "statistic": "single", "unit": "s", "value": 10.914186422014609 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[94].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.809833692245984 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[94].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[94].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[94].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[94].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3937084161303398 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[95].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[95].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[95].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40509 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[95].ttft_s", "statistic": "single", "unit": "s", "value": 15.794497117050923 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[95].decode_tps", "statistic": "single", "unit": "tok/s", "value": 48.13479652321282 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[95].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[95].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[95].dropped_messages", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[95].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3937084161303398 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[96].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[96].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[96].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60635 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[96].ttft_s", "statistic": "single", "unit": "s", "value": 10.799664471996948 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[96].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.414335916750352 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[96].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[96].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[96].dropped_messages", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[96].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3937200933285878 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[97].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[97].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[97].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40511 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[97].ttft_s", "statistic": "single", "unit": "s", "value": 15.847376855963375 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[97].decode_tps", "statistic": "single", "unit": "tok/s", "value": 48.782373506623905 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[97].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[97].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[97].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[97].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3937006317163109 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[98].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[98].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[98].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60642 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[98].ttft_s", "statistic": "single", "unit": "s", "value": 10.90455930295866 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[98].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.556913420397631 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[98].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[98].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300024 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[98].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[98].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3936928476101033 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[99].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[99].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[99].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40519 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[99].ttft_s", "statistic": "single", "unit": "s", "value": 15.880740459950175 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[99].decode_tps", "statistic": "single", "unit": "tok/s", "value": 48.90636345547181 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[99].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[99].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320028 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[99].dropped_messages", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[99].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3936694971382252 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[100].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[100].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[100].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60648 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[100].ttft_s", "statistic": "single", "unit": "s", "value": 10.906314558000304 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[100].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.083225669278503 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[100].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[100].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340031 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[100].dropped_messages", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[100].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.39368117202791675 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[101].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[101].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[101].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40515 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[101].ttft_s", "statistic": "single", "unit": "s", "value": 15.868034295039251 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[101].decode_tps", "statistic": "single", "unit": "tok/s", "value": 48.97773975240791 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[101].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[101].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[101].dropped_messages", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[101].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3936772803210787 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[102].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[102].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[102].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60645 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[102].ttft_s", "statistic": "single", "unit": "s", "value": 7.105793178023305 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[102].decode_tps", "statistic": "single", "unit": "tok/s", "value": 63.6065395649521 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[102].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[102].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380034 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[102].dropped_messages", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[102].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6566740868991673 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[103].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[103].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[103].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40512 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[103].ttft_s", "statistic": "single", "unit": "s", "value": 9.644749034021515 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[103].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.17498317809215 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[103].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[103].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400034 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[103].dropped_messages", "statistic": "single", "unit": "count", "value": 36 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[103].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[104].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[104].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[104].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60639 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[104].ttft_s", "statistic": "single", "unit": "s", "value": 7.097965908003971 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[104].decode_tps", "statistic": "single", "unit": "tok/s", "value": 63.22574295675253 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[104].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[104].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420034 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[104].dropped_messages", "statistic": "single", "unit": "count", "value": 36 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[104].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6567390623196293 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 21.282559739018325 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 20.323367278964724 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 90 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 14.051166075485526 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 90 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 105 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "105/105" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "de282722e0c5d46bc4f15406dad310be", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Startup, inference completion, concurrent decode and retrieval correctness are separate gates.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "27dacbc1ca0c47da84e238b1", "measured_at": "2026-10-07T16:21:02.311044+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/0" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "did_not_start" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "long-context-concurrency", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "a9b2513c7b4da2c5c126534490dcbefe", "standard": "pre-standard", "system": { "driver": null, "hardware": "unrecorded hardware", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp CUDA", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "2e4b0560b64e49cdbf82d029", "measured_at": "2026-10-07T01:16:53.994768+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 618.6476233049762 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 373.11016031197505 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 611.9812251609983 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 124.21525838202797 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 502.70329995598877 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.2509986830890031 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 2012.7639974073704 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 96 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[0].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 124.21477087703533 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 250.39529000100447 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 613.468780162977 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.29195191296591716 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 998.4852350816865 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 107 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[1].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 250.39455859200098 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 373.11016031197505 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 615.9772851929883 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.4488050824227521 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 670.0862817323168 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 110 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[2].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 373.10770606796723 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 494.3464972859947 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 617.6381200430333 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8759719240035261 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 505.7505239191733 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[3].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 494.3394946529879 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 611.9812251609983 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 618.6371582249994 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.37636661184747 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 408.5354088015143 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 110 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[4].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 611.9801222000387 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 1.0659124377819043 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.8599402631790904 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 36.21346664428711 } ], "model": { "artifact_sha256": [], "publisher_repo": "nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16", "publisher_revision": "a9904d24bcc1d289a1950fa9d2b978c47cf903b9", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 611.9812251609983 } ], "reasons": [ "Observed decode overlap was 4 of 5 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "04e5839fffd3856049c11c680e5ae05a807f32bbf3dec72f7a00578188171f70", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.2509986830890031, "effective_prefill_tps": 2012.7639974073704, "first_real_token_from_dispatch_s": 124.21477087703533, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 96, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": true, "total_s": 502.70329995598877, "ttft_s": 124.21525838202797, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.29195191296591716, "effective_prefill_tps": 998.4852350816865, "first_real_token_from_dispatch_s": 250.39455859200098, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 107, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": true, "total_s": 613.468780162977, "ttft_s": 250.39529000100447, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.4488050824227521, "effective_prefill_tps": 670.0862817323168, "first_real_token_from_dispatch_s": 373.10770606796723, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 110, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": true, "total_s": 615.9772851929883, "ttft_s": 373.11016031197505, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.8759719240035261, "effective_prefill_tps": 505.7505239191733, "first_real_token_from_dispatch_s": 494.3394946529879, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": true, "total_s": 617.6381200430333, "ttft_s": 494.3464972859947, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 16.37636661184747, "effective_prefill_tps": 408.5354088015143, "first_real_token_from_dispatch_s": 611.9801222000387, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 110, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": true, "total_s": 618.6371582249994, "ttft_s": 611.9812251609983, "user_number": 0, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "shared-record-groups-v2", "kv_dtype": "q8_0", "max_output_tokens": 512, "scenario": "shared", "thinking": false }, "setup_complete": false, "setup_id": "9e5118825b2267598d5691416d3ab69e", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "shared", "input_tokens": { "max": 250016, "median": 250016, "min": 250016 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "2e4b4a0f0c664129b9f26772", "measured_at": "2026-10-07T16:53:13.193159+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 4.539323047967628 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 46 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20195 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 3.9555229140096344 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.50299762972088 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20785 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 4.539323047967628 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.19540743110302 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21305 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 4.30424079799559 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.96384016319801 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21708 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 4.385879574983846 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.7792883822723 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[3].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21704 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[4].ttft_s", "statistic": "single", "unit": "s", "value": 4.376569457992446 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.6988996918389 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[4].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[4].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21707 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[5].ttft_s", "statistic": "single", "unit": "s", "value": 4.376646193035413 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.51165423503687 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[5].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[5].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21705 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[6].ttft_s", "statistic": "single", "unit": "s", "value": 4.392211236990988 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.95151946499746 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[6].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[6].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21704 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[7].ttft_s", "statistic": "single", "unit": "s", "value": 4.387277970963623 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.69386146760482 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[7].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[7].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21705 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[8].ttft_s", "statistic": "single", "unit": "s", "value": 4.380680230970029 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.04091424798308 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[8].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[8].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21709 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[9].ttft_s", "statistic": "single", "unit": "s", "value": 4.394420352997258 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.70210860553797 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[9].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[9].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20201 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[10].ttft_s", "statistic": "single", "unit": "s", "value": 3.94859991496196 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.61882767130179 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[10].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[10].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20775 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[11].ttft_s", "statistic": "single", "unit": "s", "value": 4.142345861997455 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.14098541067695 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[11].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[11].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21293 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[12].ttft_s", "statistic": "single", "unit": "s", "value": 4.286889323964715 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.68314940177518 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[12].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[12].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21688 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[13].ttft_s", "statistic": "single", "unit": "s", "value": 4.388289898983203 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.54950312822223 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[13].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[13].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21689 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[14].ttft_s", "statistic": "single", "unit": "s", "value": 4.385697540012188 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.43312487755212 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[14].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[14].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21685 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[15].ttft_s", "statistic": "single", "unit": "s", "value": 4.39087418303825 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.37237881007896 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[15].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[15].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21689 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[16].ttft_s", "statistic": "single", "unit": "s", "value": 4.393713969038799 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.2564137524893 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140010 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[16].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[16].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21687 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[17].ttft_s", "statistic": "single", "unit": "s", "value": 4.400073775963392 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.14974371166053 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[17].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[17].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21687 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[18].ttft_s", "statistic": "single", "unit": "s", "value": 4.39249016798567 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.35804795292144 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[18].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[18].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21689 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[19].ttft_s", "statistic": "single", "unit": "s", "value": 4.400507715006825 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.45832718794969 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[19].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[19].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[20].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[20].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20196 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[20].ttft_s", "statistic": "single", "unit": "s", "value": 3.9730362870031968 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[20].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.02586098716846 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[20].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[20].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[20].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[21].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[21].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20787 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[21].ttft_s", "statistic": "single", "unit": "s", "value": 4.151398827962112 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[21].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.10865409696598 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[21].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[21].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[21].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[22].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[22].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21303 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[22].ttft_s", "statistic": "single", "unit": "s", "value": 4.309348979964852 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[22].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.82972408509129 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[22].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[22].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[22].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[23].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[23].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21684 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[23].ttft_s", "statistic": "single", "unit": "s", "value": 4.378025665006135 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[23].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.97231629026767 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[23].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[23].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[23].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[24].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[24].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21687 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[24].ttft_s", "statistic": "single", "unit": "s", "value": 4.4040342650259845 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[24].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.34907507630615 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[24].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[24].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[24].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[25].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[25].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21684 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[25].ttft_s", "statistic": "single", "unit": "s", "value": 4.382892731984612 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[25].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.3810772717866 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[25].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[25].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[25].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[26].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[26].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21684 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[26].ttft_s", "statistic": "single", "unit": "s", "value": 4.385672633012291 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[26].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.4224044646221 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[26].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[26].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[26].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[27].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[27].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21684 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[27].ttft_s", "statistic": "single", "unit": "s", "value": 4.399861300014891 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[27].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.32124153754167 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[27].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[27].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[27].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[28].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[28].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21684 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[28].ttft_s", "statistic": "single", "unit": "s", "value": 4.393564586993307 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[28].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.66423921700445 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[28].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[28].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[28].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[29].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[29].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21689 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[29].ttft_s", "statistic": "single", "unit": "s", "value": 4.401297337026335 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[29].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.04405702244894 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[29].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[29].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[29].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[30].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[30].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20199 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[30].ttft_s", "statistic": "single", "unit": "s", "value": 3.968470279010944 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[30].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.80880463801341 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[30].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[30].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[30].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[31].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[31].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20780 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[31].ttft_s", "statistic": "single", "unit": "s", "value": 4.1495314740459435 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[31].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.09066137017444 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[31].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[31].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[31].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[32].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[32].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21296 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[32].ttft_s", "statistic": "single", "unit": "s", "value": 4.3078223040211014 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[32].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.92187422503783 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[32].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[32].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[32].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[33].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[33].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21677 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[33].ttft_s", "statistic": "single", "unit": "s", "value": 4.3799438729765825 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[33].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.92909039436054 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[33].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[33].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[33].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[34].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[34].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21676 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[34].ttft_s", "statistic": "single", "unit": "s", "value": 4.383536062028725 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[34].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.37853080780006 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[34].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[34].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[34].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[35].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[35].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21677 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[35].ttft_s", "statistic": "single", "unit": "s", "value": 4.383952980977483 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[35].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.2771913675185 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[35].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[35].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[35].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[36].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[36].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21678 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[36].ttft_s", "statistic": "single", "unit": "s", "value": 4.3783611170365475 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[36].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.5827652489781 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[36].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[36].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[36].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[37].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[37].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21679 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[37].ttft_s", "statistic": "single", "unit": "s", "value": 4.3880440890206955 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[37].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.6539642239269 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[37].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[37].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[37].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[38].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[38].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21677 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[38].ttft_s", "statistic": "single", "unit": "s", "value": 4.377198994043283 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[38].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.42301623069558 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[38].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[38].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[38].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[39].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[39].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21677 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[39].ttft_s", "statistic": "single", "unit": "s", "value": 4.381422781036235 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[39].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.43472631272036 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[39].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[39].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[39].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[40].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[40].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20201 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[40].ttft_s", "statistic": "single", "unit": "s", "value": 3.9497837299713865 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[40].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.54133277809201 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[40].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[40].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[40].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[41].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[41].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20778 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[41].ttft_s", "statistic": "single", "unit": "s", "value": 4.159808931988664 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[41].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.71220264353359 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[41].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[41].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[41].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[42].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[42].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21291 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[42].ttft_s", "statistic": "single", "unit": "s", "value": 4.290103844017722 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[42].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.61031880151405 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[42].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[42].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[42].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[43].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[43].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21675 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[43].ttft_s", "statistic": "single", "unit": "s", "value": 4.395612638967577 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[43].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.59029659358616 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[43].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[43].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[43].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[44].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[44].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21675 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[44].ttft_s", "statistic": "single", "unit": "s", "value": 4.382242912019137 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[44].decode_tps", "statistic": "single", "unit": "tok/s", "value": 70.62710273054765 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[44].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[44].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[44].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[45].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[45].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21676 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[45].ttft_s", "statistic": "single", "unit": "s", "value": 4.393965268973261 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[45].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.3046974636271 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[45].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[45].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[45].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[46].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[46].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21676 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[46].ttft_s", "statistic": "single", "unit": "s", "value": 4.375926693028305 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[46].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.08002337464879 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[46].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[46].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[46].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[47].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[47].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21675 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[47].ttft_s", "statistic": "single", "unit": "s", "value": 4.386098383984063 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[47].decode_tps", "statistic": "single", "unit": "tok/s", "value": 70.80961416769564 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[47].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[47].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[47].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[48].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[48].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21676 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[48].ttft_s", "statistic": "single", "unit": "s", "value": 4.388295154029038 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[48].decode_tps", "statistic": "single", "unit": "tok/s", "value": 70.89806677199108 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[48].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[48].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[48].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[49].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[49].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21681 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[49].ttft_s", "statistic": "single", "unit": "s", "value": 4.409137164009735 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[49].decode_tps", "statistic": "single", "unit": "tok/s", "value": 70.94315517640555 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[49].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[49].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[49].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 4.385672633012291 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 3.9730362870031968 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 41 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 3.9555229140096344 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 45 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 50 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "46/50" } ], "reasons": [ "46/50 requests passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "47a03fd9bbfe612c42f992e302c0e2fe", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Operator-supplied runtime manifest; the endpoint runner cannot attest hardware or model pins. Bounded task correctness is not customer acceptance.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "2f65fb170ba9426d809523e3", "measured_at": "2026-10-07T06:49:08.449660+00:00", "metrics": [ { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[0].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[0].passed", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[0].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[1].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[1].passed", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[1].total", "statistic": "single", "unit": "count", "value": 16 } ], "model": { "artifact_sha256": [], "publisher_repo": "ornith-ai/Ornith-1.5-35B-A3B", "publisher_revision": "10fbf86fed7ecee4a061f8b499a618f46001cac1", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "structured-tasks", "observed": "9/32" } ], "reasons": [ "9/32 tasks passed." ], "status": "partial" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 9, "skipped": null, "suite": "structured-tasks", "total": 32, "version": null } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": { "context_tokens": 262144, "kv_dtype": "fp8", "speculative_decoding": "with multi-token prediction" }, "setup_complete": false, "setup_id": "93c28044c43d500e1fb8a412f09eab39", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Finite lab cohort; configured 262144 context differs from measured inputs. Aggregate uses first-output to final-completion window. Whole-host memory sampled during speed only; missing counters remain unassessed. Broader coding quality and delivery acceptance unassessed." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "30b57437e5be4894a7242b78", "measured_at": "2026-10-07T06:36:28.636857+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 82.9124321937561 }, { "basis": null, "n": 8, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 60.9859663489915 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": 8, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 66.80226031900384 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "preemptions", "scope": "machine_total", "source_field": "preemptions", "statistic": "single", "unit": "count", "value": 0.0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "mtp_acceptance_rate", "scope": "machine_total", "source_field": "mtp_acceptance_rate", "statistic": "single", "unit": "ratio", "value": 0.7118572035699108 }, { "basis": null, "n": null, "name": "correctness_probes_passed", "scope": "machine_total", "source_field": "correctness_probes_passed", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 168.68998703405228 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 49.40151786197942 }, { "basis": null, "n": null, "name": "correctness_probes_total", "scope": "machine_total", "source_field": "correctness_probes_total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": 8, "name": "decode_tps", "scope": "per_person", "source_field": "per_request_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 24.46224988798614 }, { "basis": null, "n": null, "name": "forced_output_tokens", "scope": "machine_total", "source_field": "forced_output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 93.76654434204102 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 60.98568568500923 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[0].total_s", "statistic": "single", "unit": "s", "value": 82.47436325799208 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 23.77996497292448 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 66.80226031900384 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[1].total_s", "statistic": "single", "unit": "s", "value": 82.91228163399501 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 31.719386958506952 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40100 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 60.986967779987026 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[2].total_s", "statistic": "single", "unit": "s", "value": 82.25111935299356 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 24.03105518908554 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[2].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40101 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 58.67831981094787 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[3].total_s", "statistic": "single", "unit": "s", "value": 82.68674574594479 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 21.284194198467578 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 60.98624701297376 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[4].total_s", "statistic": "single", "unit": "s", "value": 82.56184770801337 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 23.684160975294777 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[5].ttft_s", "statistic": "single", "unit": "s", "value": 60.266259978991 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[5].total_s", "statistic": "single", "unit": "s", "value": 81.96014392800862 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 23.555025978791594 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[5].user_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[5].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[5].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[5].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[6].ttft_s", "statistic": "single", "unit": "s", "value": 60.26662777096499 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[6].total_s", "statistic": "single", "unit": "s", "value": 81.49501774797682 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 24.071538187934205 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[6].user_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40100 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[6].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[6].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[6].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[7].ttft_s", "statistic": "single", "unit": "s", "value": 60.98640789196361 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[7].total_s", "statistic": "single", "unit": "s", "value": 82.66405165096512 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 23.572672642883997 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[7].user_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40106 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[7].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[7].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[7].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 8, "name": "decode_tps", "scope": "per_person", "source_field": "per_person_decode_median_tps", "statistic": "median", "unit": "tok/s", "value": 23.732062974109628 } ], "model": { "artifact_sha256": [ "648fe83171485f4be03ea58d86089e088c94c2a3dc88922233f9002f30d8bc8b", "8303cef7d375021d69dcafc60d3fad5f8186995b18b0db5e2a609aadeb8a4f51", "cb5c5978721a1e87406ae68f1cd5a855326e40ff8bc13832644c392115336987" ], "publisher_repo": "ornith-ai/Ornith-1.5-35B-A3B", "publisher_revision": "10fbf86fed7ecee4a061f8b499a618f46001cac1", "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "8/8" }, { "met": false, "name": "hidden-facts", "observed": "21/24" }, { "met": false, "name": "correctness-probes", "observed": "7/8" } ], "reasons": [ "21/24 Hidden facts found.", "7/8 probes passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 21, "skipped": null, "suite": "hidden-facts", "total": 24, "version": "1" }, { "passed": 7, "skipped": null, "suite": "correctness-probes", "total": 8, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "e2097959b1ca0142f2b93edc3ec411e474c0a6622800dd247af7a71bd9c39e83", "harness": { "name": "keep-team-load-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 23.77996497292448, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 82.47436325799208, "ttft_s": 60.98568568500923, "user_number": 0 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 31.719386958506952, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40100, "total_s": 82.91228163399501, "ttft_s": 66.80226031900384, "user_number": 1 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 24.03105518908554, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40101, "total_s": 82.25111935299356, "ttft_s": 60.986967779987026, "user_number": 2 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 21.284194198467578, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 82.68674574594479, "ttft_s": 58.67831981094787, "user_number": 3 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 23.684160975294777, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 82.56184770801337, "ttft_s": 60.98624701297376, "user_number": 4 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 23.555025978791594, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 81.96014392800862, "ttft_s": 60.266259978991, "user_number": 5 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 24.071538187934205, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40100, "total_s": 81.49501774797682, "ttft_s": 60.26662777096499, "user_number": 6 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 23.572672642883997, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40106, "total_s": 82.66405165096512, "ttft_s": 60.98640789196361, "user_number": 7 } ], "scenario": { "id": "team-load", "version": "1" }, "settings": { "concurrency": 8, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "e682723b72283ed6b9af1479c6a4f677", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": { "max": 40106, "median": 40103.0, "min": 40100 }, "output_tokens": 512, "repetitions": null, "users": 8, "warmup": null } }, { "caveats": [ "Finite lab cohort; configured 262144 context differs from measured inputs. Aggregate uses first-output to final-completion window. Whole-host memory sampled during speed only; missing counters remain unassessed. Broader coding quality and delivery acceptance unassessed." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "335a73d93a99422cbfdc3047", "measured_at": "2026-10-07T06:33:31.753429+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 16.38048481941223 }, { "basis": null, "n": 1, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 10.385699263017159 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 1, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 10.385699263017159 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "preemptions", "scope": "machine_total", "source_field": "preemptions", "statistic": "single", "unit": "count", "value": 0.0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "mtp_acceptance_rate", "scope": "machine_total", "source_field": "mtp_acceptance_rate", "statistic": "single", "unit": "ratio", "value": 0.7352941176470589 }, { "basis": null, "n": null, "name": "correctness_probes_passed", "scope": "machine_total", "source_field": "correctness_probes_passed", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 85.24074730811023 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 31.256706113682153 }, { "basis": null, "n": null, "name": "correctness_probes_total", "scope": "machine_total", "source_field": "correctness_probes_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "single", "unit": "tok/s", "value": 85.24076758835105 }, { "basis": null, "n": 1, "name": "decode_tps", "scope": "per_person", "source_field": "per_request_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 85.24076758835105 }, { "basis": null, "n": null, "name": "forced_output_tokens", "scope": "machine_total", "source_field": "forced_output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 93.02033233642578 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 10.385699263017159 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[0].total_s", "statistic": "single", "unit": "s", "value": 16.380483384011313 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 85.24076758835105 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40106 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 1, "name": "decode_tps", "scope": "per_person", "source_field": "per_person_decode_median_tps", "statistic": "median", "unit": "tok/s", "value": 85.24076758835105 } ], "model": { "artifact_sha256": [ "648fe83171485f4be03ea58d86089e088c94c2a3dc88922233f9002f30d8bc8b", "8303cef7d375021d69dcafc60d3fad5f8186995b18b0db5e2a609aadeb8a4f51", "cb5c5978721a1e87406ae68f1cd5a855326e40ff8bc13832644c392115336987" ], "publisher_repo": "ornith-ai/Ornith-1.5-35B-A3B", "publisher_revision": "10fbf86fed7ecee4a061f8b499a618f46001cac1", "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "1/1" }, { "met": true, "name": "hidden-facts", "observed": "3/3" }, { "met": true, "name": "correctness-probes", "observed": "1/1" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 3, "skipped": null, "suite": "hidden-facts", "total": 3, "version": "1" }, { "passed": 1, "skipped": null, "suite": "correctness-probes", "total": 1, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "e2097959b1ca0142f2b93edc3ec411e474c0a6622800dd247af7a71bd9c39e83", "harness": { "name": "keep-team-load-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 85.24076758835105, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40106, "total_s": 16.380483384011313, "ttft_s": 10.385699263017159, "user_number": 0 } ], "scenario": { "id": "team-load", "version": "1" }, "settings": { "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "e682723b72283ed6b9af1479c6a4f677", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": { "max": 40106, "median": 40106, "min": 40106 }, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Finite lab cohort; configured 262144 context differs from measured inputs. Aggregate uses first-output to final-completion window. Whole-host memory sampled during speed only; missing counters remain unassessed. Broader coding quality and delivery acceptance unassessed." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "34822c9e0d3840f6a0fb632d", "measured_at": "2026-10-07T04:04:27.356047+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 17.233130931854248 }, { "basis": null, "n": 1, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 9.77921269997023 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 1, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 9.77921269997023 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "preemptions", "scope": "machine_total", "source_field": "preemptions", "statistic": "single", "unit": "count", "value": 0.0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "correctness_probes_passed", "scope": "machine_total", "source_field": "correctness_probes_passed", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 68.55454876389933 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 29.710213543007644 }, { "basis": null, "n": null, "name": "correctness_probes_total", "scope": "machine_total", "source_field": "correctness_probes_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "single", "unit": "tok/s", "value": 68.55455903337788 }, { "basis": null, "n": 1, "name": "decode_tps", "scope": "per_person", "source_field": "per_request_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 68.55455903337788 }, { "basis": null, "n": null, "name": "forced_output_tokens", "scope": "machine_total", "source_field": "forced_output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 92.44107818603516 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 9.77921269997023 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[0].total_s", "statistic": "single", "unit": "s", "value": 17.233129801985342 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 68.55455903337788 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40101 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 1, "name": "decode_tps", "scope": "per_person", "source_field": "per_person_decode_median_tps", "statistic": "median", "unit": "tok/s", "value": 68.55455903337788 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "1/1" }, { "met": true, "name": "hidden-facts", "observed": "3/3" }, { "met": true, "name": "correctness-probes", "observed": "1/1" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 3, "skipped": null, "suite": "hidden-facts", "total": 3, "version": "1" }, { "passed": 1, "skipped": null, "suite": "correctness-probes", "total": 1, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "e2097959b1ca0142f2b93edc3ec411e474c0a6622800dd247af7a71bd9c39e83", "harness": { "name": "keep-team-load-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 68.55455903337788, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40101, "total_s": 17.233129801985342, "ttft_s": 9.77921269997023, "user_number": 0 } ], "scenario": { "id": "team-load", "version": "1" }, "settings": { "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "plain", "thinking": false }, "setup_complete": false, "setup_id": "a4b96c03f67c006a12a64bbc7734ee71", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": { "max": 40101, "median": 40101, "min": 40101 }, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Missing named-source control: unknown source then original source omitted after a roll. Lab retrieval and rolling-history evidence only. No summaries. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "378ddde74da24365a62aae75", "measured_at": "2026-10-07T20:16:27.316123+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 21.467948145000264 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20212 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 21.726420501014218 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.89743936615319 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20322 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 21.467948145000264 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.859493897641727 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20320 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 21.453515857981984 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.812506003747986 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60010 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 21.460732001491124 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 21.726420501014218 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 21.726420501014218 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 3 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "3/3" } ], "reasons": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "9fbdb045015b3a6790f7fe36b7e05c70", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Startup, inference completion, concurrent decode and retrieval correctness are separate gates.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "3e0057ecae4142018eb3cd9e", "measured_at": "2026-10-07T07:58:56.359877+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/0" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "did_not_start" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "long-context-concurrency", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "a9b2513c7b4da2c5c126534490dcbefe", "standard": "pre-standard", "system": { "driver": null, "hardware": "unrecorded hardware", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "46082918320749d08ca01dfa", "measured_at": "2026-10-07T01:42:51.578298+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 149.12364590604557 }, { "basis": null, "n": 1, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 146.82410303596407 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 1, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 146.82410303596407 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 146.82410303596407 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 149.0816663199803 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 43.40963582011203 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 1702.792626213154 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 99 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[0].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 146.82387985097012 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 43.40954739745563 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.6638786183002415 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 99.40583038330078 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "1/1" }, { "met": true, "name": "hidden-facts", "observed": "3/3" }, { "met": true, "name": "first-token-300-s-reference", "observed": 146.82410303596407 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 3, "skipped": null, "suite": "hidden-facts", "total": 3, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 43.40963582011203, "effective_prefill_tps": 1702.792626213154, "first_real_token_from_dispatch_s": 146.82387985097012, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 99, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 149.0816663199803, "ttft_s": 146.82410303596407, "user_number": 0, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "1455fb8523cb693bc00de6e3284af4dd", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Startup, inference completion, concurrent decode and retrieval correctness are separate gates.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "463d876d79f14e2e9ded4648", "measured_at": "2026-10-07T05:34:02.037785+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "ornith-ai/Ornith-1.5-35B-A3B", "publisher_revision": "10fbf86fed7ecee4a061f8b499a618f46001cac1", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/0" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "did_not_start" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "long-context-concurrency", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "e2ae3bd60e9f88bc8d7d8d847a21461e", "standard": "pre-standard", "system": { "driver": null, "hardware": "unrecorded hardware", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Finite lab cohort; configured 262144 context differs from measured inputs. Aggregate uses first-output to final-completion window. Whole-host memory sampled during speed only; missing counters remain unassessed. Broader coding quality and delivery acceptance unassessed." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "4bbd5d175cd0483f96a7ff42", "measured_at": "2026-10-07T05:35:59.080771+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 17.657169103622437 }, { "basis": null, "n": 1, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 10.060184687026776 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 1, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 10.060184687026776 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "preemptions", "scope": "machine_total", "source_field": "preemptions", "statistic": "single", "unit": "count", "value": 0.0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "correctness_probes_passed", "scope": "machine_total", "source_field": "correctness_probes_passed", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 67.2635316865753 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 28.99672065183774 }, { "basis": null, "n": null, "name": "correctness_probes_total", "scope": "machine_total", "source_field": "correctness_probes_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.26354344885185 }, { "basis": null, "n": 1, "name": "decode_tps", "scope": "per_person", "source_field": "per_request_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 67.26354344885185 }, { "basis": null, "n": null, "name": "forced_output_tokens", "scope": "machine_total", "source_field": "forced_output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 92.80904388427734 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 10.060184687026776 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[0].total_s", "statistic": "single", "unit": "s", "value": 17.657167744997423 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.26354344885185 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 1, "name": "decode_tps", "scope": "per_person", "source_field": "per_person_decode_median_tps", "statistic": "median", "unit": "tok/s", "value": 67.26354344885185 } ], "model": { "artifact_sha256": [ "648fe83171485f4be03ea58d86089e088c94c2a3dc88922233f9002f30d8bc8b", "8303cef7d375021d69dcafc60d3fad5f8186995b18b0db5e2a609aadeb8a4f51", "cb5c5978721a1e87406ae68f1cd5a855326e40ff8bc13832644c392115336987" ], "publisher_repo": "ornith-ai/Ornith-1.5-35B-A3B", "publisher_revision": "10fbf86fed7ecee4a061f8b499a618f46001cac1", "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "1/1" }, { "met": true, "name": "hidden-facts", "observed": "3/3" }, { "met": true, "name": "correctness-probes", "observed": "1/1" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 3, "skipped": null, "suite": "hidden-facts", "total": 3, "version": "1" }, { "passed": 1, "skipped": null, "suite": "correctness-probes", "total": 1, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "e2097959b1ca0142f2b93edc3ec411e474c0a6622800dd247af7a71bd9c39e83", "harness": { "name": "keep-team-load-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 67.26354344885185, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 17.657167744997423, "ttft_s": 10.060184687026776, "user_number": 0 } ], "scenario": { "id": "team-load", "version": "1" }, "settings": { "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "plain", "thinking": false }, "setup_complete": false, "setup_id": "a167e4bafd3fd96d0b47cf9d6645473d", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": { "max": 40104, "median": 40104, "min": 40104 }, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Startup, inference completion, concurrent decode and retrieval correctness are separate gates.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "50a609d1dc83489398596228", "measured_at": "2026-10-07T06:29:17.328846+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "ornith-ai/Ornith-1.5-35B-A3B", "publisher_revision": "10fbf86fed7ecee4a061f8b499a618f46001cac1", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/0" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "did_not_start" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "long-context-concurrency", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "e2ae3bd60e9f88bc8d7d8d847a21461e", "standard": "pre-standard", "system": { "driver": null, "hardware": "unrecorded hardware", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "5341cef0c186439485cace00", "measured_at": "2026-10-07T18:52:12.255379+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 4.083014553005341 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20195 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 3.95114751404617 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.65898160042705 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20359 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 4.053044608968776 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.38053395255615 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20361 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 4.074882480024826 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.43189615761443 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20363 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 4.053318022051826 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.086635001141 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[3].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20359 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[4].ttft_s", "statistic": "single", "unit": "s", "value": 4.072083327977452 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.96964838837266 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[4].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[4].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20359 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[5].ttft_s", "statistic": "single", "unit": "s", "value": 4.07284309505485 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.85440734924607 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[5].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[5].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20361 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[6].ttft_s", "statistic": "single", "unit": "s", "value": 4.050678492989391 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.04485974910433 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[6].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[6].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20360 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[7].ttft_s", "statistic": "single", "unit": "s", "value": 4.063443353981711 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.81271282463513 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[7].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[7].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20361 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[8].ttft_s", "statistic": "single", "unit": "s", "value": 4.0567754820222035 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.7190220041769 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[8].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[8].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20366 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[9].ttft_s", "statistic": "single", "unit": "s", "value": 4.083014553005341 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.8361649734873 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[9].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[9].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20201 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[10].ttft_s", "statistic": "single", "unit": "s", "value": 3.9537245079991408 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.76535849819088 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[10].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[10].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20362 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[11].ttft_s", "statistic": "single", "unit": "s", "value": 4.034611585957464 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.79432220680746 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[11].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[11].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20362 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[12].ttft_s", "statistic": "single", "unit": "s", "value": 4.060538418008946 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.96163248985837 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[12].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[12].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20363 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[13].ttft_s", "statistic": "single", "unit": "s", "value": 4.060249769012444 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.13426732089462 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[13].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[13].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20362 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[14].ttft_s", "statistic": "single", "unit": "s", "value": 4.057297352002934 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.91054183868077 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[14].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[14].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20363 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[15].ttft_s", "statistic": "single", "unit": "s", "value": 4.0450106769567356 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.231265615231 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[15].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[15].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20366 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[16].ttft_s", "statistic": "single", "unit": "s", "value": 4.065628582029603 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.40246416019797 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140010 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[16].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[16].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20363 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[17].ttft_s", "statistic": "single", "unit": "s", "value": 4.063338704989292 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.20578150196006 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[17].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[17].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20365 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[18].ttft_s", "statistic": "single", "unit": "s", "value": 4.063476900977548 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.7483562234025 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[18].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[18].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20364 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[19].ttft_s", "statistic": "single", "unit": "s", "value": 4.074222469993401 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.61599700363568 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[19].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[19].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 4.061938561499119 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 3.9537245079991408 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 3.9524360110226553 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 20 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "19/20" } ], "reasons": [ "19/20 requests passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 2, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "47a03fd9bbfe612c42f992e302c0e2fe", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Finite lab cohort; configured 262144 context differs from measured inputs. Aggregate uses first-output to final-completion window. Whole-host memory sampled during speed only; missing counters remain unassessed. Broader coding quality and delivery acceptance unassessed." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "5366b32d63594303bd6c627d", "measured_at": "2026-10-07T05:38:56.618035+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 85.29438710212708 }, { "basis": null, "n": 8, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 54.295966231497005 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": 8, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 63.35083819000283 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "preemptions", "scope": "machine_total", "source_field": "preemptions", "statistic": "single", "unit": "count", "value": 0.0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "correctness_probes_passed", "scope": "machine_total", "source_field": "correctness_probes_passed", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 125.33113262551221 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 48.02191725811526 }, { "basis": null, "n": null, "name": "correctness_probes_total", "scope": "machine_total", "source_field": "correctness_probes_total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": 8, "name": "decode_tps", "scope": "per_person", "source_field": "per_request_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 18.18877871935077 }, { "basis": null, "n": null, "name": "forced_output_tokens", "scope": "machine_total", "source_field": "forced_output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 94.6588134765625 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 54.29679778503487 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[0].total_s", "statistic": "single", "unit": "s", "value": 84.91481842304347 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.68951778566818 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40096 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 54.295235718018375 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[1].total_s", "statistic": "single", "unit": "s", "value": 84.91270488203736 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.68981839297536 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 63.26338509900961 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[2].total_s", "statistic": "single", "unit": "s", "value": 85.28199579700595 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 23.2076404369373 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[2].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 54.29382729099598 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[3].total_s", "statistic": "single", "unit": "s", "value": 84.9107472699834 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.690117763338126 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 63.35083819000283 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[4].total_s", "statistic": "single", "unit": "s", "value": 85.29077736200998 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 23.290857645219802 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40097 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[5].ttft_s", "statistic": "single", "unit": "s", "value": 52.676520195032936 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[5].total_s", "statistic": "single", "unit": "s", "value": 84.83710313600022 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 15.889015473941242 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[5].user_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[5].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[5].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[5].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[6].ttft_s", "statistic": "single", "unit": "s", "value": 53.644833061960526 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[6].total_s", "statistic": "single", "unit": "s", "value": 84.87352550000651 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.363157087468945 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[6].user_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[6].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[6].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[6].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[7].ttft_s", "statistic": "single", "unit": "s", "value": 54.296696744975634 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[7].total_s", "statistic": "single", "unit": "s", "value": 84.91363982699113 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.69010516925719 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[7].user_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40097 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[7].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[7].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[7].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 8, "name": "decode_tps", "scope": "per_person", "source_field": "per_person_decode_median_tps", "statistic": "median", "unit": "tok/s", "value": 16.689961781116274 } ], "model": { "artifact_sha256": [ "648fe83171485f4be03ea58d86089e088c94c2a3dc88922233f9002f30d8bc8b", "8303cef7d375021d69dcafc60d3fad5f8186995b18b0db5e2a609aadeb8a4f51", "cb5c5978721a1e87406ae68f1cd5a855326e40ff8bc13832644c392115336987" ], "publisher_repo": "ornith-ai/Ornith-1.5-35B-A3B", "publisher_revision": "10fbf86fed7ecee4a061f8b499a618f46001cac1", "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "8/8" }, { "met": false, "name": "hidden-facts", "observed": "18/24" }, { "met": false, "name": "correctness-probes", "observed": "6/8" } ], "reasons": [ "18/24 Hidden facts found.", "6/8 probes passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 18, "skipped": null, "suite": "hidden-facts", "total": 24, "version": "1" }, { "passed": 6, "skipped": null, "suite": "correctness-probes", "total": 8, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "e2097959b1ca0142f2b93edc3ec411e474c0a6622800dd247af7a71bd9c39e83", "harness": { "name": "keep-team-load-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 16.68951778566818, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40096, "total_s": 84.91481842304347, "ttft_s": 54.29679778503487, "user_number": 0 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 16.68981839297536, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 84.91270488203736, "ttft_s": 54.295235718018375, "user_number": 1 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 23.2076404369373, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 85.28199579700595, "ttft_s": 63.26338509900961, "user_number": 2 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 16.690117763338126, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 84.9107472699834, "ttft_s": 54.29382729099598, "user_number": 3 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 23.290857645219802, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40097, "total_s": 85.29077736200998, "ttft_s": 63.35083819000283, "user_number": 4 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 15.889015473941242, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 84.83710313600022, "ttft_s": 52.676520195032936, "user_number": 5 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 16.363157087468945, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 84.87352550000651, "ttft_s": 53.644833061960526, "user_number": 6 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 16.69010516925719, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40097, "total_s": 84.91363982699113, "ttft_s": 54.296696744975634, "user_number": 7 } ], "scenario": { "id": "team-load", "version": "1" }, "settings": { "concurrency": 8, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "plain", "thinking": false }, "setup_complete": false, "setup_id": "a167e4bafd3fd96d0b47cf9d6645473d", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": { "max": 40105, "median": 40102.5, "min": 40096 }, "output_tokens": 512, "repetitions": null, "users": 8, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "5eaafaf613c8480ab606b56e", "measured_at": "2026-10-07T02:07:57.409450+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 8.253126458032057 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 3.365631385007873 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 4.23519023595145 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 1.3545920900069177 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 7.890181193011813 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.37189828087634 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[0].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 1.346116903994698 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 4.2347962519852445 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 7.932456801994704 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 28.125891653233 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 105 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[1].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 4.221264690975659 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 3.365631385007873 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 7.996740407019388 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 23.104616948431225 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[2].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 3.3646409069770016 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 4.23519023595145 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 8.110177016991656 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 28.387193612689295 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 111 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[3].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 4.230429955990985 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 2.4487347250105813 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 8.243046777963173 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 20.19221590600745 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 118 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[4].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 2.4374950799974613 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 79.12132143482646 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 66.64141192998835 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 100.55206680297852 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "4/5" }, { "met": false, "name": "hidden-facts", "observed": "14/15" }, { "met": true, "name": "first-token-300-s-reference", "observed": 4.23519023595145 } ], "reasons": [ "One or more requests did not meet completion, exact retrieval, synthesis-presence, reasoning-off or token-accounting requirements." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 14, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "77fda79ee1961c35a8f8a77e7d5a01d5f9ce414975e00e640a166e71e5f9093e", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 16.37189828087634, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 1.346116903994698, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 7.890181193011813, "ttft_s": 1.3545920900069177, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 28.125891653233, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 4.221264690975659, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 105, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 7.932456801994704, "ttft_s": 4.2347962519852445, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 23.104616948431225, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 3.3646409069770016, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 7.996740407019388, "ttft_s": 3.365631385007873, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 28.387193612689295, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 4.230429955990985, "first_token_within_300_seconds": true, "needle_correct": 2, "needle_total": 3, "output_tokens": 111, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 8.110177016991656, "ttft_s": 4.23519023595145, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 20.19221590600745, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 2.4374950799974613, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 118, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 8.243046777963173, "ttft_s": 2.4487347250105813, "user_number": 0, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "warm", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "shared-record-groups-v2", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "shared", "thinking": false }, "setup_complete": false, "setup_id": "1455fb8523cb693bc00de6e3284af4dd", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": 250000, "document": "shared", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab retrieval and rolling-history evidence only. No summaries. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "5f1b8f8e534b4032a6bb6205", "measured_at": "2026-10-07T08:41:29.987232+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 276.9408023909782 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 59 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20113 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 4.618168626038823 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.22161198092034 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40238 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 5.750403939047828 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 68.10853804162626 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.46881057706645457 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60364 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 6.936156009964179 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 63.89760780663465 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6597309654761115 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80492 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 8.793969994003419 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 60.39141043333744 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[3].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7291159369875267 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100620 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[4].ttft_s", "statistic": "single", "unit": "s", "value": 9.998475282045547 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 57.3246430945164 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[4].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[4].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7915722520373684 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120744 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[5].ttft_s", "statistic": "single", "unit": "s", "value": 10.969639275979716 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 55.230106100159404 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[5].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[5].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8332339495130193 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 140868 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[6].ttft_s", "statistic": "single", "unit": "s", "value": 15.045422182010952 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 23.654115853468635 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140015 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[6].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[6].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7421174249012349 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 160993 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[7].ttft_s", "statistic": "single", "unit": "s", "value": 18.294226290017832 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 34.9141736494623 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160018 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[7].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[7].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7915879862450058 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181122 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[8].ttft_s", "statistic": "single", "unit": "s", "value": 21.938967138004955 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 37.88337205359319 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[8].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[8].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8245407715369453 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201252 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[9].ttft_s", "statistic": "single", "unit": "s", "value": 24.98725502099842 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 34.65078745777543 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[9].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[9].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8480679749140199 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221374 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[10].ttft_s", "statistic": "single", "unit": "s", "value": 26.370293710031547 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 34.00472285047838 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[10].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[10].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8722538649308381 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241504 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[11].ttft_s", "statistic": "single", "unit": "s", "value": 32.42389117501443 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 25.35181609858997 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[11].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[11].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8386872366745212 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181126 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[12].ttft_s", "statistic": "single", "unit": "s", "value": 119.6631423549843 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.0898485424172075 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[12].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[12].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6348141198629043 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201257 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[13].ttft_s", "statistic": "single", "unit": "s", "value": 21.212238314037677 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.662981628243096 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[13].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[13].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8411879371631693 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[14].ttft_s", "statistic": "single", "unit": "s", "value": 35.191805834008846 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.741973086726519 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[14].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[14].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8365441486437317 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241508 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[15].ttft_s", "statistic": "single", "unit": "s", "value": 59.34387544798665 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.418249920390636 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[15].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[15].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8534072386136687 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181129 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[16].ttft_s", "statistic": "single", "unit": "s", "value": 216.08381266403012 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.0994652802838267 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[16].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[16].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6190763072994842 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201255 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[17].ttft_s", "statistic": "single", "unit": "s", "value": 33.661038674996234 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.850790385108196 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[17].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[17].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8828432274828159 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[18].ttft_s", "statistic": "single", "unit": "s", "value": 65.85493068298092 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.964734673010654 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[18].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[18].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8947719052394516 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241505 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[19].ttft_s", "statistic": "single", "unit": "s", "value": 67.59845666901674 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.3453966869678626 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[19].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[19].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7352456589096446 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[20].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[20].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181116 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[20].ttft_s", "statistic": "single", "unit": "s", "value": 275.56003946700366 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[20].decode_tps", "statistic": "single", "unit": "tok/s", "value": 8.097949994422367 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[20].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[20].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[20].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6585290649162007 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[21].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[21].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20120 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[21].ttft_s", "statistic": "single", "unit": "s", "value": 11.534934628987685 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[21].decode_tps", "statistic": "single", "unit": "tok/s", "value": 42.001809200707264 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[21].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[21].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[21].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[22].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[22].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40243 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[22].ttft_s", "statistic": "single", "unit": "s", "value": 18.29874032997759 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[22].decode_tps", "statistic": "single", "unit": "tok/s", "value": 38.47391805018253 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[22].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40005 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[22].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[22].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7915879862450058 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[23].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[23].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60370 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[23].ttft_s", "statistic": "single", "unit": "s", "value": 21.918794583994895 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[23].decode_tps", "statistic": "single", "unit": "tok/s", "value": 37.887463791191344 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[23].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[23].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[23].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8245407715369453 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[24].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[24].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80499 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[24].ttft_s", "statistic": "single", "unit": "s", "value": 24.54568315000506 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[24].decode_tps", "statistic": "single", "unit": "tok/s", "value": 29.304702438262954 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[24].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[24].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[24].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8480679749140199 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[25].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[25].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100624 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[25].ttft_s", "statistic": "single", "unit": "s", "value": 26.51707644900307 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[25].decode_tps", "statistic": "single", "unit": "tok/s", "value": 33.61232034586498 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[25].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[25].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[25].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8722538649308381 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[26].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[26].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120747 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[26].ttft_s", "statistic": "single", "unit": "s", "value": 29.08639305300312 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[26].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.957114014693733 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[26].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[26].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[26].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8386872366745212 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[27].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[27].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 140875 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[27].ttft_s", "statistic": "single", "unit": "s", "value": 19.780554877012037 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[27].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.5706198443544297 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[27].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140015 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[27].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[27].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3818830797770582 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[28].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[28].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 161002 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[28].ttft_s", "statistic": "single", "unit": "s", "value": 30.12820964801358 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[28].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.9641252072748985 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[28].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160019 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[28].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[28].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.814260030536558 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[29].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[29].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181131 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[29].ttft_s", "statistic": "single", "unit": "s", "value": 24.17035638599191 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[29].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.572241594319842 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[29].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[29].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[29].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8245066293446845 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[30].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[30].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201259 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[30].ttft_s", "statistic": "single", "unit": "s", "value": 25.83791077398928 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[30].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.525681836657026 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[30].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[30].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[30].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8517754786960302 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[31].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[31].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[31].ttft_s", "statistic": "single", "unit": "s", "value": 41.402984159998596 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[31].decode_tps", "statistic": "single", "unit": "tok/s", "value": 19.878199031096354 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[31].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[31].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[31].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8331810465003934 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[32].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[32].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241516 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[32].ttft_s", "statistic": "single", "unit": "s", "value": 60.022816899989266 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[32].decode_tps", "statistic": "single", "unit": "tok/s", "value": 19.780660751413773 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[32].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[32].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[32].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8534072386136687 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[33].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[33].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181140 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[33].ttft_s", "statistic": "single", "unit": "s", "value": 215.30731486994773 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[33].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.1224647444063804 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[33].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[33].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[33].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6190763072994842 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[34].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[34].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201268 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[34].ttft_s", "statistic": "single", "unit": "s", "value": 33.455539491027594 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[34].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.8449074854844079 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[34].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[34].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[34].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8828432274828159 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[35].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[35].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221391 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[35].ttft_s", "statistic": "single", "unit": "s", "value": 63.178587372007314 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[35].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.318523661127239 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[35].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[35].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[35].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8947719052394516 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[36].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[36].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241517 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[36].ttft_s", "statistic": "single", "unit": "s", "value": 66.49909634998767 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[36].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.226531038575835 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[36].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[36].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[36].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6231003090708505 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[37].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[37].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181141 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[37].ttft_s", "statistic": "single", "unit": "s", "value": 276.9408023909782 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[37].decode_tps", "statistic": "single", "unit": "tok/s", "value": 19.099557391533203 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[37].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[37].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[37].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7589662402842924 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[38].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[38].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201270 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[38].ttft_s", "statistic": "single", "unit": "s", "value": 65.3564359479933 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[38].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.962230463653537 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[38].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360027 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[38].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[38].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9064720256189053 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[39].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[39].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221403 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[39].ttft_s", "statistic": "single", "unit": "s", "value": 55.9022045910242 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[39].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.3412339151349495 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[39].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[39].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[39].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7242202823142041 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[40].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[40].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241531 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[40].ttft_s", "statistic": "single", "unit": "s", "value": 40.937298393051606 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[40].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8946588557224519 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[40].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[40].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[40].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5207034512455667 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[41].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[41].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181148 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[41].ttft_s", "statistic": "single", "unit": "s", "value": 228.8798174689873 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[41].decode_tps", "statistic": "single", "unit": "tok/s", "value": 17.572935399423447 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[41].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420032 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[41].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[41].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7372764896138158 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[42].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[42].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20118 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[42].ttft_s", "statistic": "single", "unit": "s", "value": 27.368014677020255 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[42].decode_tps", "statistic": "single", "unit": "tok/s", "value": 27.866679839929333 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[42].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[42].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[42].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[43].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[43].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40246 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[43].ttft_s", "statistic": "single", "unit": "s", "value": 20.187084902019706 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[43].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.49325938479366 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[43].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[43].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[43].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5327635599876923 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[44].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[44].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60372 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[44].ttft_s", "statistic": "single", "unit": "s", "value": 29.64320204500109 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[44].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.9167051675033997 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[44].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[44].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[44].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6596435433644736 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[45].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[45].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80497 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[45].ttft_s", "statistic": "single", "unit": "s", "value": 23.41692826099461 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[45].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.3865753591256564 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[45].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[45].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[45].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8245066293446845 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[46].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[46].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100625 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[46].ttft_s", "statistic": "single", "unit": "s", "value": 25.99032236705534 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[46].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.448297540418438 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[46].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[46].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[46].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8517754786960302 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[47].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[47].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120754 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[47].ttft_s", "statistic": "single", "unit": "s", "value": 40.51031227398198 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[47].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.362915771665058 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[47].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120017 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[47].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[47].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8331810465003934 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[48].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[48].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 140885 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[48].ttft_s", "statistic": "single", "unit": "s", "value": 60.09026365500176 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[48].decode_tps", "statistic": "single", "unit": "tok/s", "value": 19.030258925299076 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[48].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140019 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[48].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[48].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8534072386136687 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[49].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[49].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 161011 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[49].ttft_s", "statistic": "single", "unit": "s", "value": 33.26254644297296 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[49].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.3863198920140178 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[49].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160022 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[49].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[49].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.35977306399739817 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[50].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[50].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[50].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181140 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[50].ttft_s", "statistic": "single", "unit": "s", "value": 53.18440152500989 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[50].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0855546083315646 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[50].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[50].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[50].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[50].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8215728004725744 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[51].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[51].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[51].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201268 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[51].ttft_s", "statistic": "single", "unit": "s", "value": 35.07577661599498 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[51].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.172667349064797 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[51].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[51].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[51].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[51].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8737831369103524 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[52].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[52].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[52].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221392 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[52].ttft_s", "statistic": "single", "unit": "s", "value": 39.29667099897051 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[52].decode_tps", "statistic": "single", "unit": "tok/s", "value": 22.1729085519154 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[52].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[52].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[52].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[52].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8721870510150425 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[53].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[53].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[53].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241516 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[53].ttft_s", "statistic": "single", "unit": "s", "value": 69.02672828099458 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[53].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.490096924161392 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[53].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[53].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[53].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[53].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8947719052394516 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[54].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[54].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[54].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181133 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[54].ttft_s", "statistic": "single", "unit": "s", "value": 215.4399368049926 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[54].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.6779046375397965 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[54].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[54].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260027 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[54].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[54].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6720918774115221 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[55].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[55].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[55].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201261 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[55].ttft_s", "statistic": "single", "unit": "s", "value": 46.060043659992516 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[55].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.45220138513206 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[55].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[55].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[55].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[55].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8956330337223803 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[56].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[56].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[56].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221389 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[56].ttft_s", "statistic": "single", "unit": "s", "value": 64.9460785970441 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[56].decode_tps", "statistic": "single", "unit": "tok/s", "value": 15.447522758212218 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[56].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[56].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[56].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[56].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9064720256189053 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[57].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[57].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[57].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241516 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[57].ttft_s", "statistic": "single", "unit": "s", "value": 57.47292300296249 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[57].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.3610904871414267 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[57].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[57].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320030 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[57].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[57].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7242202823142041 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[58].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[58].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[58].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181128 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[58].ttft_s", "statistic": "single", "unit": "s", "value": 251.12354657403193 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[58].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.6631063185753192 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[58].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[58].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340030 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[58].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[58].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4889101095305102 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[59].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[59].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[59].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201258 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[59].ttft_s", "statistic": "single", "unit": "s", "value": 62.15007812198019 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[59].decode_tps", "statistic": "single", "unit": "tok/s", "value": 14.548587917123275 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[59].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[59].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360030 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[59].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[59].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9025270245652056 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[60].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[60].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[60].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221387 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[60].ttft_s", "statistic": "single", "unit": "s", "value": 48.99225218099309 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[60].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.395897335288746 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[60].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[60].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380031 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[60].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[60].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.909642462606744 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[61].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[61].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[61].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241516 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[61].ttft_s", "statistic": "single", "unit": "s", "value": 40.71945388900349 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[61].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.3863378363920034 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[61].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[61].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400031 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[61].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[61].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.662658039184986 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[62].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[62].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[62].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181142 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[62].ttft_s", "statistic": "single", "unit": "s", "value": 211.0221390450024 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[62].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.078092558279616 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[62].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[62].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420031 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[62].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[62].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.319829098954757 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[63].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[63].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[63].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20119 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[63].ttft_s", "statistic": "single", "unit": "s", "value": 17.668054998037405 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[63].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.8118741385985295 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[63].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[63].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[63].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[63].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8071477594929239 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[64].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[64].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[64].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40247 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[64].ttft_s", "statistic": "single", "unit": "s", "value": 39.02865377702983 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[64].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.244570710129142 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[64].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[64].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[64].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[64].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8365441486437317 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[65].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[65].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[65].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60377 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[65].ttft_s", "statistic": "single", "unit": "s", "value": 59.296773840964306 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[65].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.151273288777826 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[65].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[65].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60005 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[65].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[65].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8534072386136687 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[66].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[66].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[66].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80500 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[66].ttft_s", "statistic": "single", "unit": "s", "value": 32.4696967570344 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[66].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.3879783889259878 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[66].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[66].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80005 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[66].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[66].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.35977306399739817 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[67].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[67].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[67].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100629 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[67].ttft_s", "statistic": "single", "unit": "s", "value": 52.04594863799866 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[67].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9821154705718683 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[67].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[67].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[67].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[67].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8215728004725744 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[68].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[68].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[68].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120753 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[68].ttft_s", "statistic": "single", "unit": "s", "value": 31.7028984290082 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[68].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.065266391914872 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[68].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[68].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[68].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[68].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8737831369103524 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[69].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[69].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[69].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 140880 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[69].ttft_s", "statistic": "single", "unit": "s", "value": 39.82661087799352 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[69].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.375057469331559 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[69].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[69].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[69].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[69].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8941910579866207 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[70].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[70].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[70].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 161011 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[70].ttft_s", "statistic": "single", "unit": "s", "value": 63.287249733984936 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[70].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.25126770568787 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[70].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[70].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[70].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[70].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8608829296730086 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[71].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[71].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[71].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181138 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[71].ttft_s", "statistic": "single", "unit": "s", "value": 66.80216403899249 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[71].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.230185166907539 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[71].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[71].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180016 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[71].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[71].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6231003090708505 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[72].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[72].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[72].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201271 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[72].ttft_s", "statistic": "single", "unit": "s", "value": 51.48094061698066 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[72].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8633684046893519 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[72].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[72].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200018 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[72].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[72].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6796243949762436 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[73].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[73].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[73].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221403 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[73].ttft_s", "statistic": "single", "unit": "s", "value": 52.65402664698195 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[73].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0300310426818617 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[73].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[73].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220018 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[73].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[73].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9025419718736396 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[74].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[74].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[74].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241528 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[74].ttft_s", "statistic": "single", "unit": "s", "value": 67.30384509603027 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[74].decode_tps", "statistic": "single", "unit": "tok/s", "value": 15.873521300163576 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[74].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[74].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240018 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[74].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[74].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9064720256189053 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[75].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[75].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[75].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181153 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[75].ttft_s", "statistic": "single", "unit": "s", "value": 192.69289196201134 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[75].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.843429315329734 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[75].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[75].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[75].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[75].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5156671451472136 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[76].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[76].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[76].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201282 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[76].ttft_s", "statistic": "single", "unit": "s", "value": 51.59853641997324 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[76].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.6732081449868124 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[76].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[76].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[76].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[76].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.895539591220278 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[77].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[77].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[77].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221413 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[77].ttft_s", "statistic": "single", "unit": "s", "value": 59.577260018966626 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[77].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.047442245087584 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[77].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[77].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[77].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[77].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9002700515261517 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[78].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[78].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[78].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241544 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[78].ttft_s", "statistic": "single", "unit": "s", "value": 52.309726169973146 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[78].decode_tps", "statistic": "single", "unit": "tok/s", "value": 20.900984978553378 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[78].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[78].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[78].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[78].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.909642462606744 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[79].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[79].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[79].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181165 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[79].ttft_s", "statistic": "single", "unit": "s", "value": 155.45922738296213 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[79].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9512449177740487 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[79].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[79].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[79].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[79].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.428780591872301 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[80].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[80].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[80].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201296 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[80].ttft_s", "statistic": "single", "unit": "s", "value": 44.990673171007074 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[80].decode_tps", "statistic": "single", "unit": "tok/s", "value": 24.207594791864206 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[80].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[80].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[80].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[80].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8954773070503139 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[81].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[81].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[81].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221428 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[81].ttft_s", "statistic": "single", "unit": "s", "value": 32.9928958559758 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[81].decode_tps", "statistic": "single", "unit": "tok/s", "value": 29.211942289848633 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[81].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[81].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[81].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[81].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9024224072672218 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[82].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[82].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[82].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241559 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[82].ttft_s", "statistic": "single", "unit": "s", "value": 35.59710663399892 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[82].decode_tps", "statistic": "single", "unit": "tok/s", "value": 26.968346942484963 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[82].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[82].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400026 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[82].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[82].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9099599769321983 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[83].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[83].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[83].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181168 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[83].ttft_s", "statistic": "single", "unit": "s", "value": 121.23448496998753 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[83].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.70629827382646 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[83].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[83].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420027 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[83].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[83].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.36443783543002106 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[84].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[84].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[84].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20118 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[84].ttft_s", "statistic": "single", "unit": "s", "value": 37.69683398201596 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[84].decode_tps", "statistic": "single", "unit": "tok/s", "value": 19.75243397182322 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[84].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[84].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[84].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[84].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8331890965254802 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[85].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[85].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[85].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40245 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[85].ttft_s", "statistic": "single", "unit": "s", "value": 60.04868661996443 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[85].decode_tps", "statistic": "single", "unit": "tok/s", "value": 19.204095305831594 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[85].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[85].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[85].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[85].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8534072386136687 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[86].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[86].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[86].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60372 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[86].ttft_s", "statistic": "single", "unit": "s", "value": 33.310133725986816 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[86].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.3428602066364117 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[86].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[86].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[86].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[86].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5052092088028493 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[87].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[87].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[87].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80502 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[87].ttft_s", "statistic": "single", "unit": "s", "value": 52.86910520901438 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[87].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.1254662891471872 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[87].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[87].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[87].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[87].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7290253658294203 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[88].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[88].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[88].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100631 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[88].ttft_s", "statistic": "single", "unit": "s", "value": 34.961029901052825 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[88].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.296423581057103 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[88].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[88].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[88].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[88].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8737831369103524 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[89].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[89].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[89].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120762 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[89].ttft_s", "statistic": "single", "unit": "s", "value": 39.32973897800548 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[89].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.9371778574699983 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[89].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[89].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[89].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[89].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.891524106925076 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[90].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[90].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[90].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 140893 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[90].ttft_s", "statistic": "single", "unit": "s", "value": 56.76970623701345 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[90].decode_tps", "statistic": "single", "unit": "tok/s", "value": 15.79079713203034 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[90].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[90].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[90].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[90].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8479626383141817 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[91].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[91].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[91].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 161021 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[91].ttft_s", "statistic": "single", "unit": "s", "value": 66.21555742801866 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[91].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.250851705467765 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[91].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[91].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160017 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[91].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[91].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6231003090708505 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[92].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[92].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[92].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181153 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[92].ttft_s", "statistic": "single", "unit": "s", "value": 51.69549452699721 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[92].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8863400525878786 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[92].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[92].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[92].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[92].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6025072529344441 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[93].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[93].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[93].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201281 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[93].ttft_s", "statistic": "single", "unit": "s", "value": 54.78207608603407 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[93].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9914406436876828 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[93].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[93].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[93].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[93].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.900284480202582 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[94].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[94].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[94].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221409 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[94].ttft_s", "statistic": "single", "unit": "s", "value": 65.36430979700526 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[94].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.397684373383246 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[94].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[94].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[94].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[94].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9064720256189053 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[95].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[95].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[95].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241537 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[95].ttft_s", "statistic": "single", "unit": "s", "value": 56.99289693601895 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[95].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.3920121622757367 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[95].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[95].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[95].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[95].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7242202823142041 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[96].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[96].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[96].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181150 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[96].ttft_s", "statistic": "single", "unit": "s", "value": 246.4088098150096 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[96].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.5625153884514407 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[96].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[96].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[96].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[96].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.40592189114403304 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[97].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[97].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[97].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201284 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[97].ttft_s", "statistic": "single", "unit": "s", "value": 62.7972112930147 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[97].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.189232096186991 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[97].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[97].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[97].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[97].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9002700515261517 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[98].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[98].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[98].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221417 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[98].ttft_s", "statistic": "single", "unit": "s", "value": 51.884084224002436 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[98].decode_tps", "statistic": "single", "unit": "tok/s", "value": 19.271660986251483 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[98].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[98].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300024 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[98].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[98].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.909642462606744 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[99].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[99].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[99].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241553 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[99].ttft_s", "statistic": "single", "unit": "s", "value": 41.324446061044 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[99].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.3951749897812427 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[99].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[99].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320028 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[99].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[99].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.662658039184986 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[100].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[100].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[100].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181163 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[100].ttft_s", "statistic": "single", "unit": "s", "value": 212.3616311540245 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[100].decode_tps", "statistic": "single", "unit": "tok/s", "value": 23.85100270911458 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[100].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[100].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340031 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[100].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[100].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.319829098954757 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[101].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[101].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[101].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201292 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[101].ttft_s", "statistic": "single", "unit": "s", "value": 33.331794086960144 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[101].decode_tps", "statistic": "single", "unit": "tok/s", "value": 29.18647040005001 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[101].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[101].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[101].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[101].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9024224072672218 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[102].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[102].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[102].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221424 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[102].ttft_s", "statistic": "single", "unit": "s", "value": 35.19575730198994 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[102].decode_tps", "statistic": "single", "unit": "tok/s", "value": 24.117941040549717 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[102].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[102].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380034 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[102].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[102].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9099599769321983 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[103].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[103].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[103].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241552 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[103].ttft_s", "statistic": "single", "unit": "s", "value": 24.324803489027545 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[103].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.474693724667383 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[103].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[103].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400034 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[103].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[103].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5206283118849356 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[104].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[104].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[104].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181169 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[104].ttft_s", "statistic": "single", "unit": "s", "value": 128.955080610991 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[104].decode_tps", "statistic": "single", "unit": "tok/s", "value": 48.529753151255655 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[104].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[104].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420034 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[104].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[104].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 215.30731486994773 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 69.02672828099458 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 39.16266238800017 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 105 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "59/105" } ], "reasons": [ "59/105 requests passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "b33ca7489f90d7eb7ee7ac3f45f81993", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "60ac465f5ee74a1c86991f67", "measured_at": "2026-10-07T17:04:05.941286+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 4.418206533999182 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20195 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 3.963336065004114 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.56373240373794 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20785 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 4.162940542970318 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.97864597426442 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21318 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 4.3391759599908255 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.209337386707 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21675 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 4.402229733997956 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.73016115506518 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[3].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21671 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[4].ttft_s", "statistic": "single", "unit": "s", "value": 4.418206533999182 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.92593688159698 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[4].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[4].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21674 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[5].ttft_s", "statistic": "single", "unit": "s", "value": 4.410259619995486 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.05355303822742 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[5].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[5].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21672 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[6].ttft_s", "statistic": "single", "unit": "s", "value": 4.406980736006517 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.70678485774351 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[6].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[6].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21671 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[7].ttft_s", "statistic": "single", "unit": "s", "value": 4.416321319004055 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.55998785051241 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[7].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[7].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21672 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[8].ttft_s", "statistic": "single", "unit": "s", "value": 4.409330889990088 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.5521360880813 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[8].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[8].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21676 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[9].ttft_s", "statistic": "single", "unit": "s", "value": 4.404636758030392 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.07271248716567 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[9].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[9].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20201 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[10].ttft_s", "statistic": "single", "unit": "s", "value": 3.961893827014137 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.88359451358555 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[10].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[10].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20775 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[11].ttft_s", "statistic": "single", "unit": "s", "value": 4.161715116002597 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.54548318575023 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[11].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[11].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21306 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[12].ttft_s", "statistic": "single", "unit": "s", "value": 4.327246999018826 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.4666232698078 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[12].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[12].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21655 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[13].ttft_s", "statistic": "single", "unit": "s", "value": 4.404681844986044 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.33094900139606 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[13].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[13].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21656 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[14].ttft_s", "statistic": "single", "unit": "s", "value": 4.3998075039708056 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 70.9426650502763 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[14].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[14].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21652 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[15].ttft_s", "statistic": "single", "unit": "s", "value": 4.405373340996448 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.50572227060506 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[15].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[15].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21656 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[16].ttft_s", "statistic": "single", "unit": "s", "value": 4.402984750980977 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.40493102910536 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140010 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[16].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[16].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21654 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[17].ttft_s", "statistic": "single", "unit": "s", "value": 4.399688861973118 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.49318205235434 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[17].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[17].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21654 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[18].ttft_s", "statistic": "single", "unit": "s", "value": 4.393129948992282 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.69908625712743 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[18].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[18].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21656 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[19].ttft_s", "statistic": "single", "unit": "s", "value": 4.410305983037688 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 70.90282259358969 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[19].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[19].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 4.403810754505685 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 3.963336065004114 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 3.9626149460091256 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 20 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "19/20" } ], "reasons": [ "19/20 requests passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 2, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "47a03fd9bbfe612c42f992e302c0e2fe", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Startup, inference completion, concurrent decode and retrieval correctness are separate gates.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "6177be2b391e4efcbd235f00", "measured_at": "2026-10-07T01:39:20.760860+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/0" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "did_not_start" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "long-context-concurrency", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "a9b2513c7b4da2c5c126534490dcbefe", "standard": "pre-standard", "system": { "driver": null, "hardware": "unrecorded hardware", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Finite lab cohort; configured 262144 context differs from measured inputs. Aggregate uses first-output to final-completion window. Whole-host memory sampled during speed only; missing counters remain unassessed. Broader coding quality and delivery acceptance unassessed." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "64ac3c82131c4ec28814f973", "measured_at": "2026-10-07T06:48:36.191129+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 198.4367320537567 }, { "basis": null, "n": 20, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 118.86346217850223 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 60 }, { "basis": null, "n": 20, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 169.94332427403424 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 39 }, { "basis": null, "n": null, "name": "preemptions", "scope": "machine_total", "source_field": "preemptions", "statistic": "single", "unit": "count", "value": 0.0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "mtp_acceptance_rate", "scope": "machine_total", "source_field": "mtp_acceptance_rate", "statistic": "single", "unit": "ratio", "value": 0.7056269828018034 }, { "basis": null, "n": null, "name": "correctness_probes_passed", "scope": "machine_total", "source_field": "correctness_probes_passed", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 73.35129734348108 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 51.60334930947146 }, { "basis": null, "n": null, "name": "correctness_probes_total", "scope": "machine_total", "source_field": "correctness_probes_total", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": 20, "name": "decode_tps", "scope": "per_person", "source_field": "per_request_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 8.281334057708495 }, { "basis": null, "n": null, "name": "forced_output_tokens", "scope": "machine_total", "source_field": "forced_output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 93.93363189697266 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 110.63064979796764 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[0].total_s", "statistic": "single", "unit": "s", "value": 196.10757276898948 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.978221749667308 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40101 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 169.94332427403424 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[1].total_s", "statistic": "single", "unit": "s", "value": 198.43398820300354 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 17.935699963819214 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40099 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 95.21379447705112 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[2].total_s", "statistic": "single", "unit": "s", "value": 194.6932208520011 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.136740516315195 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[2].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 169.81731461902382 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[3].total_s", "statistic": "single", "unit": "s", "value": 198.15822209999897 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 18.03047415976454 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40098 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 59.106821419962216 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[4].total_s", "statistic": "single", "unit": "s", "value": 195.20962499896996 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.75451487083706 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[5].ttft_s", "statistic": "single", "unit": "s", "value": 62.74957149999682 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[5].total_s", "statistic": "single", "unit": "s", "value": 194.31795417703688 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.8839118457080066 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[5].user_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[5].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[5].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[5].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[6].ttft_s", "statistic": "single", "unit": "s", "value": 128.69373133900808 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[6].total_s", "statistic": "single", "unit": "s", "value": 197.16893128899392 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.462555792071195 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[6].user_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[6].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[6].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[6].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[7].ttft_s", "statistic": "single", "unit": "s", "value": 142.71225799800595 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[7].total_s", "statistic": "single", "unit": "s", "value": 196.04307077103294 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.581702836121924 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[7].user_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[7].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[7].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[7].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[8].ttft_s", "statistic": "single", "unit": "s", "value": 62.748756188957486 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[8].total_s", "statistic": "single", "unit": "s", "value": 192.09367063594982 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.9506771656601636 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[8].user_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[8].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[8].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[8].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[9].ttft_s", "statistic": "single", "unit": "s", "value": 95.21450705401367 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[9].total_s", "statistic": "single", "unit": "s", "value": 195.75162407400785 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.082699953474651 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[9].user_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40101 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[9].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[9].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[9].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[10].ttft_s", "statistic": "single", "unit": "s", "value": 154.83195108000655 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[10].total_s", "statistic": "single", "unit": "s", "value": 197.60637446300825 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.946391314840454 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[10].user_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[10].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[10].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[10].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[11].ttft_s", "statistic": "single", "unit": "s", "value": 93.52838986803545 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[11].total_s", "statistic": "single", "unit": "s", "value": 196.2355097749969 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.975312329494741 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[11].user_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40100 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[11].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[11].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[11].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[12].ttft_s", "statistic": "single", "unit": "s", "value": 62.750055051990785 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[12].total_s", "statistic": "single", "unit": "s", "value": 194.95759075297974 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.8651352004300126 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[12].user_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[12].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[12].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[12].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[13].ttft_s", "statistic": "single", "unit": "s", "value": 151.23978388798423 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[13].total_s", "statistic": "single", "unit": "s", "value": 197.7582752329763 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.984879028219204 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[13].user_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40097 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[13].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[13].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[13].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[14].ttft_s", "statistic": "single", "unit": "s", "value": 128.69420517899562 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[14].total_s", "statistic": "single", "unit": "s", "value": 197.12180199398426 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.467747280115916 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[14].user_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[14].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[14].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[14].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[15].ttft_s", "statistic": "single", "unit": "s", "value": 166.7397651670035 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[15].total_s", "statistic": "single", "unit": "s", "value": 196.91859792597825 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.932397753125027 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[15].user_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[15].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[15].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[15].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[16].ttft_s", "statistic": "single", "unit": "s", "value": 62.74925806099782 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[16].total_s", "statistic": "single", "unit": "s", "value": 194.01250293501653 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.892940483761754 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[16].user_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[16].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[16].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[16].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[17].ttft_s", "statistic": "single", "unit": "s", "value": 159.95084790198598 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[17].total_s", "statistic": "single", "unit": "s", "value": 197.88762173097348 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.469780068898341 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[17].user_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40106 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[17].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[17].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[17].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[18].ttft_s", "statistic": "single", "unit": "s", "value": 127.09627455903683 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[18].total_s", "statistic": "single", "unit": "s", "value": 195.67789716500556 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.450975648912791 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[18].user_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[18].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[18].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[18].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[19].ttft_s", "statistic": "single", "unit": "s", "value": 61.07461324398173 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[19].total_s", "statistic": "single", "unit": "s", "value": 194.01171275199158 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.8439231929323894 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[19].user_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40108 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[19].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[19].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[19].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 20, "name": "decode_tps", "scope": "per_person", "source_field": "per_person_decode_median_tps", "statistic": "median", "unit": "tok/s", "value": 6.7145986992900495 } ], "model": { "artifact_sha256": [ "648fe83171485f4be03ea58d86089e088c94c2a3dc88922233f9002f30d8bc8b", "8303cef7d375021d69dcafc60d3fad5f8186995b18b0db5e2a609aadeb8a4f51", "cb5c5978721a1e87406ae68f1cd5a855326e40ff8bc13832644c392115336987" ], "publisher_repo": "ornith-ai/Ornith-1.5-35B-A3B", "publisher_revision": "10fbf86fed7ecee4a061f8b499a618f46001cac1", "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "20/20" }, { "met": false, "name": "hidden-facts", "observed": "39/60" }, { "met": false, "name": "correctness-probes", "observed": "13/20" } ], "reasons": [ "39/60 Hidden facts found.", "13/20 probes passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 39, "skipped": null, "suite": "hidden-facts", "total": 60, "version": "1" }, { "passed": 13, "skipped": null, "suite": "correctness-probes", "total": 20, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "e2097959b1ca0142f2b93edc3ec411e474c0a6622800dd247af7a71bd9c39e83", "harness": { "name": "keep-team-load-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.978221749667308, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40101, "total_s": 196.10757276898948, "ttft_s": 110.63064979796764, "user_number": 0 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 17.935699963819214, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40099, "total_s": 198.43398820300354, "ttft_s": 169.94332427403424, "user_number": 1 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 5.136740516315195, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 194.6932208520011, "ttft_s": 95.21379447705112, "user_number": 2 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 18.03047415976454, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40098, "total_s": 198.15822209999897, "ttft_s": 169.81731461902382, "user_number": 3 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.75451487083706, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 195.20962499896996, "ttft_s": 59.106821419962216, "user_number": 4 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 3.8839118457080066, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 194.31795417703688, "ttft_s": 62.74957149999682, "user_number": 5 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.462555792071195, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 197.16893128899392, "ttft_s": 128.69373133900808, "user_number": 6 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 9.581702836121924, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 196.04307077103294, "ttft_s": 142.71225799800595, "user_number": 7 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 3.9506771656601636, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 192.09367063594982, "ttft_s": 62.748756188957486, "user_number": 8 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.082699953474651, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40101, "total_s": 195.75162407400785, "ttft_s": 95.21450705401367, "user_number": 9 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 11.946391314840454, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 197.60637446300825, "ttft_s": 154.83195108000655, "user_number": 10 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 4.975312329494741, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40100, "total_s": 196.2355097749969, "ttft_s": 93.52838986803545, "user_number": 11 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.8651352004300126, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 194.95759075297974, "ttft_s": 62.750055051990785, "user_number": 12 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 10.984879028219204, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40097, "total_s": 197.7582752329763, "ttft_s": 151.23978388798423, "user_number": 13 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 7.467747280115916, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 197.12180199398426, "ttft_s": 128.69420517899562, "user_number": 14 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 16.932397753125027, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 196.91859792597825, "ttft_s": 166.7397651670035, "user_number": 15 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.892940483761754, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 194.01250293501653, "ttft_s": 62.74925806099782, "user_number": 16 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 13.469780068898341, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40106, "total_s": 197.88762173097348, "ttft_s": 159.95084790198598, "user_number": 17 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.450975648912791, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 195.67789716500556, "ttft_s": 127.09627455903683, "user_number": 18 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.8439231929323894, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40108, "total_s": 194.01171275199158, "ttft_s": 61.07461324398173, "user_number": 19 } ], "scenario": { "id": "team-load", "version": "1" }, "settings": { "concurrency": 20, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "e682723b72283ed6b9af1479c6a4f677", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": { "max": 40108, "median": 40103.0, "min": 40097 }, "output_tokens": 512, "repetitions": null, "users": 20, "warmup": null } }, { "caveats": [ "Measured retrieval only, with no customer qualification threshold. Exact dense float32 array size excludes payloads and vector database overhead. One timing run; no reranking, ingestion, answer generation or GPU measurement.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Transformers CPU", "recipe": "plain", "version": "5.19.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "65a87f4af24c4e02b56e1ca4", "measured_at": "2026-10-07T14:39:45.943909+00:00", "metrics": [ { "basis": null, "n": null, "name": "passages_per_second", "scope": "machine_total", "source_field": "passages_per_second", "statistic": "single", "unit": "passages/s", "value": 12.25921414124026 }, { "basis": null, "n": null, "name": "query_seconds", "scope": "per_person", "source_field": "query_seconds", "statistic": "single", "unit": "s", "value": 2.574559000087902 }, { "basis": null, "n": null, "name": "recall_at_5", "scope": "machine_total", "source_field": "recall_at_5", "statistic": "single", "unit": "ratio", "value": 1.0 }, { "basis": null, "n": null, "name": "mrr", "scope": "machine_total", "source_field": "mrr", "statistic": "single", "unit": "ratio", "value": 0.9725 }, { "basis": null, "n": null, "name": "questions", "scope": "machine_total", "source_field": "questions", "statistic": "single", "unit": "count", "value": 100 }, { "basis": null, "n": null, "name": "passage_seconds", "scope": "machine_total", "source_field": "passage_seconds", "statistic": "single", "unit": "s", "value": 32.62851887498982 }, { "basis": null, "n": null, "name": "vector_bytes", "scope": "machine_total", "source_field": "vector_bytes", "statistic": "single", "unit": "bytes", "value": 819200 }, { "basis": null, "n": null, "name": "passages", "scope": "machine_total", "source_field": "passages", "statistic": "single", "unit": "count", "value": 400 }, { "basis": null, "n": null, "name": "index_file_bytes", "scope": "machine_total", "source_field": "index_file_bytes", "statistic": "single", "unit": "bytes", "value": 819328 }, { "basis": null, "n": null, "name": "recall_at_10", "scope": "machine_total", "source_field": "recall_at_10", "statistic": "single", "unit": "ratio", "value": 1.0 }, { "basis": null, "n": null, "name": "truncated_passages", "scope": "machine_total", "source_field": "truncated_passages", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "google/embeddinggemma-2", "publisher_revision": "914f7f89142e33e77833254d9c9b90c3cef7303b", "variant": "FP32 variant", "weights_dtype": "FP32", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-embedding-eval-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "embeddings", "version": "1" }, "settings": { "context_tokens": 262144, "corpus_sha256": "d7fb630b25a7d5128c44491d1a8f21bde479711f79520f1ce95ea8bd8a0ae1f6", "dimensions": 512, "dtype": "float32" }, "setup_complete": false, "setup_id": "1040c93fe6bb05121a008bfd66febe48", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple Silicon laptop CPU, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Finite lab cohort; configured 262144 context differs from measured inputs. Aggregate uses first-output to final-completion window. Whole-host memory sampled during speed only; missing counters remain unassessed. Broader coding quality and delivery acceptance unassessed." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "68118ad14a454699b5052087", "measured_at": "2026-10-07T06:41:57.846748+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 159.96905493736267 }, { "basis": null, "n": 16, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 95.07330801652279 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 48 }, { "basis": null, "n": 16, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 135.08585100696655 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 30 }, { "basis": null, "n": null, "name": "preemptions", "scope": "machine_total", "source_field": "preemptions", "statistic": "single", "unit": "count", "value": 0.0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "mtp_acceptance_rate", "scope": "machine_total", "source_field": "mtp_acceptance_rate", "statistic": "single", "unit": "ratio", "value": 0.7108307045215563 }, { "basis": null, "n": null, "name": "correctness_probes_passed", "scope": "machine_total", "source_field": "correctness_probes_passed", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 80.96538619271466 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 51.20990433560823 }, { "basis": null, "n": null, "name": "correctness_probes_total", "scope": "machine_total", "source_field": "correctness_probes_total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": 16, "name": "decode_tps", "scope": "per_person", "source_field": "per_request_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 10.431920443153471 }, { "basis": null, "n": null, "name": "forced_output_tokens", "scope": "machine_total", "source_field": "forced_output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 93.93132400512695 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 135.08585100696655 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[0].total_s", "statistic": "single", "unit": "s", "value": 159.96094153699232 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 20.542638805000184 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 95.0737348080147 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[1].total_s", "statistic": "single", "unit": "s", "value": 158.83388462703442 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 8.014410277429558 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 95.07288122503087 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[2].total_s", "statistic": "single", "unit": "s", "value": 157.9459811480483 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 8.127482192315544 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[2].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 62.64551535103237 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[3].total_s", "statistic": "single", "unit": "s", "value": 156.7550567430444 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.429842632761713 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 134.18076786398888 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[4].total_s", "statistic": "single", "unit": "s", "value": 159.61795602797065 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 20.088698353993358 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[5].ttft_s", "statistic": "single", "unit": "s", "value": 58.98438179696677 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[5].total_s", "statistic": "single", "unit": "s", "value": 156.21267304295907 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.2556719186511796 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[5].user_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[5].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[5].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[5].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[6].ttft_s", "statistic": "single", "unit": "s", "value": 130.57057565398281 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[6].total_s", "statistic": "single", "unit": "s", "value": 159.3935762999463 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 17.72889666404539 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[6].user_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[6].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[6].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[6].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[7].ttft_s", "statistic": "single", "unit": "s", "value": 123.22267978900345 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[7].total_s", "statistic": "single", "unit": "s", "value": 159.2675608129939 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 14.176770334181233 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[7].user_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[7].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[7].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[7].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[8].ttft_s", "statistic": "single", "unit": "s", "value": 93.43093882902758 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[8].total_s", "statistic": "single", "unit": "s", "value": 158.51897636504145 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.8509050102679545 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[8].user_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40099 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[8].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[8].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[8].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[9].ttft_s", "statistic": "single", "unit": "s", "value": 123.22135551803512 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[9].total_s", "statistic": "single", "unit": "s", "value": 159.396370794042 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 14.125771505587206 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[9].user_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[9].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[9].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[9].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[10].ttft_s", "statistic": "single", "unit": "s", "value": 60.9555434640497 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[10].total_s", "statistic": "single", "unit": "s", "value": 156.9136344169965 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.325241414510525 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[10].user_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[10].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[10].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[10].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[11].ttft_s", "statistic": "single", "unit": "s", "value": 62.648104756954126 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[11].total_s", "statistic": "single", "unit": "s", "value": 157.2439310689806 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.401929661404457 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[11].user_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[11].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[11].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[11].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[12].ttft_s", "statistic": "single", "unit": "s", "value": 122.28557898901636 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[12].total_s", "statistic": "single", "unit": "s", "value": 159.2178093570401 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.836153270678947 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[12].user_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[12].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[12].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[12].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[13].ttft_s", "statistic": "single", "unit": "s", "value": 62.64641078998102 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[13].total_s", "statistic": "single", "unit": "s", "value": 157.322281130997 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.397362582032921 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[13].user_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40101 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[13].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[13].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[13].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[14].ttft_s", "statistic": "single", "unit": "s", "value": 62.64637023000978 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[14].total_s", "statistic": "single", "unit": "s", "value": 157.53911308798706 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.385027185532999 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[14].user_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40100 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[14].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[14].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[14].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[15].ttft_s", "statistic": "single", "unit": "s", "value": 109.08666853199247 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[15].total_s", "statistic": "single", "unit": "s", "value": 159.06747198099038 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.223925282062375 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[15].user_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[15].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[15].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[15].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 16, "name": "decode_tps", "scope": "per_person", "source_field": "per_person_decode_median_tps", "statistic": "median", "unit": "tok/s", "value": 8.070946234872551 } ], "model": { "artifact_sha256": [ "648fe83171485f4be03ea58d86089e088c94c2a3dc88922233f9002f30d8bc8b", "8303cef7d375021d69dcafc60d3fad5f8186995b18b0db5e2a609aadeb8a4f51", "cb5c5978721a1e87406ae68f1cd5a855326e40ff8bc13832644c392115336987" ], "publisher_repo": "ornith-ai/Ornith-1.5-35B-A3B", "publisher_revision": "10fbf86fed7ecee4a061f8b499a618f46001cac1", "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "16/16" }, { "met": false, "name": "hidden-facts", "observed": "30/48" }, { "met": false, "name": "correctness-probes", "observed": "10/16" } ], "reasons": [ "30/48 Hidden facts found.", "10/16 probes passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 30, "skipped": null, "suite": "hidden-facts", "total": 48, "version": "1" }, { "passed": 10, "skipped": null, "suite": "correctness-probes", "total": 16, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "e2097959b1ca0142f2b93edc3ec411e474c0a6622800dd247af7a71bd9c39e83", "harness": { "name": "keep-team-load-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 20.542638805000184, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 159.96094153699232, "ttft_s": 135.08585100696655, "user_number": 0 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 8.014410277429558, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 158.83388462703442, "ttft_s": 95.0737348080147, "user_number": 1 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 8.127482192315544, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 157.9459811480483, "ttft_s": 95.07288122503087, "user_number": 2 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 5.429842632761713, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 156.7550567430444, "ttft_s": 62.64551535103237, "user_number": 3 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 20.088698353993358, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 159.61795602797065, "ttft_s": 134.18076786398888, "user_number": 4 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.2556719186511796, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 156.21267304295907, "ttft_s": 58.98438179696677, "user_number": 5 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 17.72889666404539, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 159.3935762999463, "ttft_s": 130.57057565398281, "user_number": 6 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 14.176770334181233, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 159.2675608129939, "ttft_s": 123.22267978900345, "user_number": 7 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 7.8509050102679545, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40099, "total_s": 158.51897636504145, "ttft_s": 93.43093882902758, "user_number": 8 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 14.125771505587206, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 159.396370794042, "ttft_s": 123.22135551803512, "user_number": 9 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.325241414510525, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 156.9136344169965, "ttft_s": 60.9555434640497, "user_number": 10 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 5.401929661404457, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 157.2439310689806, "ttft_s": 62.648104756954126, "user_number": 11 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 13.836153270678947, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 159.2178093570401, "ttft_s": 122.28557898901636, "user_number": 12 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 5.397362582032921, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40101, "total_s": 157.322281130997, "ttft_s": 62.64641078998102, "user_number": 13 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.385027185532999, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40100, "total_s": 157.53911308798706, "ttft_s": 62.64637023000978, "user_number": 14 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 10.223925282062375, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 159.06747198099038, "ttft_s": 109.08666853199247, "user_number": 15 } ], "scenario": { "id": "team-load", "version": "1" }, "settings": { "concurrency": 16, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "e682723b72283ed6b9af1479c6a4f677", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": { "max": 40105, "median": 40103.0, "min": 40099 }, "output_tokens": 512, "repetitions": null, "users": 16, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "68146b5a341248dd966978ee", "measured_at": "2026-10-07T05:25:56.863392+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 17.645647894998547 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 14.472924098023213 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 14.473109697981272 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 14.472991042013746 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 17.214713967987336 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 35.74394738126211 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 99 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[0].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 14.455858776986133 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 14.472924098023213 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 17.285651412035804 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 36.619262552352396 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 104 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[1].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 14.472163506026845 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 10.883061390020885 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 17.341255374019966 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.568099419915974 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 108 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[2].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 10.8780743360403 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 13.262781532015651 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 17.50440759799676 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 27.583761081244045 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 118 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[3].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 13.252208704012446 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 14.473109697981272 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 17.63643365801545 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 36.354164623327904 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 116 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[4].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 14.459321542992257 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 79.96312746722312 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 30.88580273408232 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 93.86608505249023 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "3/5" }, { "met": false, "name": "hidden-facts", "observed": "13/15" }, { "met": true, "name": "first-token-300-s-reference", "observed": 14.473109697981272 } ], "reasons": [ "One or more requests did not meet completion, exact retrieval, synthesis-presence, reasoning-off or token-accounting requirements." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 13, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "77fda79ee1961c35a8f8a77e7d5a01d5f9ce414975e00e640a166e71e5f9093e", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 35.74394738126211, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 14.455858776986133, "first_token_within_300_seconds": true, "needle_correct": 2, "needle_total": 3, "output_tokens": 99, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 17.214713967987336, "ttft_s": 14.472991042013746, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 36.619262552352396, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 14.472163506026845, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 104, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 17.285651412035804, "ttft_s": 14.472924098023213, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 16.568099419915974, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 10.8780743360403, "first_token_within_300_seconds": true, "needle_correct": 2, "needle_total": 3, "output_tokens": 108, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 17.341255374019966, "ttft_s": 10.883061390020885, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 27.583761081244045, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 13.252208704012446, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 118, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 17.50440759799676, "ttft_s": 13.262781532015651, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 36.354164623327904, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 14.459321542992257, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 116, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 17.63643365801545, "ttft_s": 14.473109697981272, "user_number": 4, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "warm", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "shared-record-groups-v2", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "shared", "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "d0a27f1110ad464348786ff3b4c49fa3", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": 250000, "document": "shared", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "6abf3f90e1c24c8a8f6c77a9", "measured_at": "2026-10-07T20:36:22.746541+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 61.348261887033004 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 50 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20208 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 21.250106174033135 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.830055161953016 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20377 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 21.54418666299898 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.793414952293974 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 34.634467445954215 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.23937019524669 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 27.158287475002 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.1380407455037416 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[3].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20377 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[4].ttft_s", "statistic": "single", "unit": "s", "value": 35.94867890601745 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.9630154045141595 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[4].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[4].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20377 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[5].ttft_s", "statistic": "single", "unit": "s", "value": 38.83586469502188 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.0228465755940377 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[5].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[5].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[6].ttft_s", "statistic": "single", "unit": "s", "value": 42.27296624798328 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.339294774161382 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[6].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[6].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20378 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[7].ttft_s", "statistic": "single", "unit": "s", "value": 61.348261887033004 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.3126622002515291 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[7].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[7].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[8].ttft_s", "statistic": "single", "unit": "s", "value": 51.6276127989986 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.111739059531205 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[8].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[8].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20384 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[9].ttft_s", "statistic": "single", "unit": "s", "value": 48.37421912100399 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9665274445664771 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[9].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[9].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20214 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[10].ttft_s", "statistic": "single", "unit": "s", "value": 35.9470733400085 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.8684579218483943 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[10].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[10].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[11].ttft_s", "statistic": "single", "unit": "s", "value": 28.924052838003263 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.069477537708321 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[11].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[11].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[12].ttft_s", "statistic": "single", "unit": "s", "value": 45.25182168697938 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.293969023746613 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[12].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[12].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[13].ttft_s", "statistic": "single", "unit": "s", "value": 40.16147531301249 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.4363252227885726 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[13].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[13].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[14].ttft_s", "statistic": "single", "unit": "s", "value": 42.64908678102074 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.3414435539836003 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[14].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[14].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[15].ttft_s", "statistic": "single", "unit": "s", "value": 59.02513823797926 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.2722554408886029 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[15].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[15].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20384 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[16].ttft_s", "statistic": "single", "unit": "s", "value": 49.54346055100905 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0284568017965048 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140010 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[16].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[16].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[17].ttft_s", "statistic": "single", "unit": "s", "value": 46.6173863550066 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0210471306132696 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[17].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[17].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[18].ttft_s", "statistic": "single", "unit": "s", "value": 35.64969206403475 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.2123033387121431 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[18].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[18].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[19].ttft_s", "statistic": "single", "unit": "s", "value": 47.82871890201932 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.500172035764838 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[19].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[19].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[20].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[20].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20209 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[20].ttft_s", "statistic": "single", "unit": "s", "value": 32.47820810001576 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[20].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.6249684869186387 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[20].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[20].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[20].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[21].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[21].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[21].ttft_s", "statistic": "single", "unit": "s", "value": 22.0166380580049 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[21].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.3639880212244564 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[21].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[21].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[21].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[22].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[22].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[22].ttft_s", "statistic": "single", "unit": "s", "value": 21.978308780991938 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[22].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9084304458367519 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[22].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[22].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[22].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[23].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[23].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20378 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[23].ttft_s", "statistic": "single", "unit": "s", "value": 42.30601453199051 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[23].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9798136402494707 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[23].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[23].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[23].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[24].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[24].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[24].ttft_s", "statistic": "single", "unit": "s", "value": 25.09237333299825 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[24].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8226102586544148 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[24].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[24].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[24].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[25].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[25].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[25].ttft_s", "statistic": "single", "unit": "s", "value": 33.255420306988526 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[25].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0717109663562956 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[25].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[25].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[25].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[26].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[26].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[26].ttft_s", "statistic": "single", "unit": "s", "value": 29.60758736002026 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[26].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.345011044570258 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[26].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[26].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[26].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[27].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[27].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[27].ttft_s", "statistic": "single", "unit": "s", "value": 37.92664998897817 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[27].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.642461531874093 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[27].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[27].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[27].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[28].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[28].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[28].ttft_s", "statistic": "single", "unit": "s", "value": 29.05132173601305 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[28].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.09581902054128 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[28].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[28].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[28].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[29].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[29].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[29].ttft_s", "statistic": "single", "unit": "s", "value": 37.131354147044476 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[29].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.4023807180965595 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[29].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[29].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[29].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[30].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[30].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20212 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[30].ttft_s", "statistic": "single", "unit": "s", "value": 58.220151921967044 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[30].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.1040753330003998 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[30].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[30].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[30].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[31].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[31].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[31].ttft_s", "statistic": "single", "unit": "s", "value": 42.11487901001237 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[31].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9156946344369653 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[31].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[31].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[31].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[32].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[32].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20385 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[32].ttft_s", "statistic": "single", "unit": "s", "value": 43.52218569803517 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[32].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.43769881169095 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[32].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[32].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[32].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[33].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[33].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[33].ttft_s", "statistic": "single", "unit": "s", "value": 38.07068127999082 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[33].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.4080594618304887 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[33].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[33].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[33].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[34].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[34].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20376 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[34].ttft_s", "statistic": "single", "unit": "s", "value": 49.753555187024176 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[34].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.515614054880994 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[34].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[34].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[34].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[35].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[35].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[35].ttft_s", "statistic": "single", "unit": "s", "value": 27.43734979000874 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[35].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.6177062347955646 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[35].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[35].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[35].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[36].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[36].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[36].ttft_s", "statistic": "single", "unit": "s", "value": 35.471352129010484 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[36].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.9971716198585494 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[36].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[36].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[36].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[37].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[37].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20385 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[37].ttft_s", "statistic": "single", "unit": "s", "value": 22.025877153035253 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[37].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.4925837025746533 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[37].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[37].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[37].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[38].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[38].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[38].ttft_s", "statistic": "single", "unit": "s", "value": 21.919768079009373 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[38].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.6409369542652925 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[38].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[38].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[38].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[39].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[39].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20384 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[39].ttft_s", "statistic": "single", "unit": "s", "value": 21.95582591398852 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[39].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.669470717148332 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[39].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[39].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[39].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[40].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[40].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20214 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[40].ttft_s", "statistic": "single", "unit": "s", "value": 42.029381842992734 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[40].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8808523325588203 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[40].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[40].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[40].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[41].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[41].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[41].ttft_s", "statistic": "single", "unit": "s", "value": 24.172552177042235 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[41].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7543257055218343 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[41].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[41].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[41].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[42].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[42].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[42].ttft_s", "statistic": "single", "unit": "s", "value": 25.85289458098123 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[42].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0878989621776975 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[42].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[42].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[42].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[43].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[43].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[43].ttft_s", "statistic": "single", "unit": "s", "value": 29.187868449022062 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[43].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9807878589878045 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[43].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[43].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[43].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[44].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[44].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[44].ttft_s", "statistic": "single", "unit": "s", "value": 22.480829048960004 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[44].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.4234675209746417 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[44].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[44].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[44].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[45].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[45].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[45].ttft_s", "statistic": "single", "unit": "s", "value": 21.99216823495226 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[45].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.4805845859238966 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[45].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[45].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[45].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[46].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[46].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[46].ttft_s", "statistic": "single", "unit": "s", "value": 21.959460643993225 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[46].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.475002761852092 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[46].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[46].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[46].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[47].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[47].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[47].ttft_s", "statistic": "single", "unit": "s", "value": 21.927982586959843 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[47].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.6110441942701255 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[47].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[47].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[47].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[48].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[48].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[48].ttft_s", "statistic": "single", "unit": "s", "value": 21.922018234967254 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[48].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.550689086841161 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[48].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[48].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[48].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[49].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[49].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20386 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[49].ttft_s", "statistic": "single", "unit": "s", "value": 22.068512653000653 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[49].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.56606581491945 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[49].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[49].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[49].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 34.634467445954215 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 58.220151921967044 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 45 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 35.9470733400085 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 45 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 50 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "50/50" } ], "reasons": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "9fbdb045015b3a6790f7fe36b7e05c70", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp CUDA", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "6fc815a108f34202abcaff2e", "measured_at": "2026-10-07T00:56:03.723718+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 120.90144624200184 }, { "basis": null, "n": 1, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 118.03834720898885 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 1, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 118.03834720898885 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 118.03834720898885 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 120.89322150498629 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 31.174752641396164 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 2118.0913314326785 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 90 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[0].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 118.03743578796275 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 31.174738788744083 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.7444079686181082 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 35.79835510253906 } ], "model": { "artifact_sha256": [], "publisher_repo": "nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16", "publisher_revision": "a9904d24bcc1d289a1950fa9d2b978c47cf903b9", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "1/1" }, { "met": true, "name": "hidden-facts", "observed": "3/3" }, { "met": true, "name": "first-token-300-s-reference", "observed": 118.03834720898885 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 3, "skipped": null, "suite": "hidden-facts", "total": 3, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "04cf656671acc024178598a26462c2de38b6f97137ea5abf44a09b83b8047c5e", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 31.174752641396164, "effective_prefill_tps": 2118.0913314326785, "first_real_token_from_dispatch_s": 118.03743578796275, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 90, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": true, "total_s": 120.89322150498629, "ttft_s": 118.03834720898885, "user_number": 0, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "q8_0", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "9e5118825b2267598d5691416d3ab69e", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250016, "median": 250016, "min": 250016 }, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Bounded fictional tasks only. Unsupported cases are separately counted, never passes. Original publisher name with variant; exact variant repository and revision retained privately. Not customer acceptance.", "Publisher revision was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "70d5791f1fec47958088dcc3", "measured_at": "2026-10-07T05:52:52.779398+00:00", "metrics": [ { "basis": null, "n": null, "name": "skipped_cases", "scope": "machine_total", "source_field": "skipped_cases", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "correctness_failures", "scope": "machine_total", "source_field": "correctness_failures", "statistic": "single", "unit": "count", "value": 50 }, { "basis": null, "n": null, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 0.057789848011452705 }, { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 130 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 71 }, { "basis": null, "n": null, "name": "api_cases_passed", "scope": "machine_total", "source_field": "api_cases_passed", "statistic": "single", "unit": "count", "value": 71 }, { "basis": null, "n": null, "name": "ttft_p95_s", "scope": "per_person", "source_field": "ttft_p95_s", "statistic": "p95", "unit": "s", "value": 0.10831664898432791 }, { "basis": null, "n": null, "name": "completion_p95_s", "scope": "per_person", "source_field": "completion_p95_s", "statistic": "p95", "unit": "s", "value": 0.7688227570033632 }, { "basis": null, "n": null, "name": "api_cases_total", "scope": "machine_total", "source_field": "api_cases_total", "statistic": "single", "unit": "count", "value": 124 }, { "basis": null, "n": null, "name": "completion_median_s", "scope": "per_person", "source_field": "completion_median_s", "statistic": "median", "unit": "s", "value": 0.23296860296977684 }, { "basis": null, "n": null, "name": "completeness_failures", "scope": "machine_total", "source_field": "completeness_failures", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[0].passed", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[0].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[0].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[1].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[1].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[1].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[2].passed", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[2].total", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[2].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[3].passed", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[3].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[3].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[4].passed", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[4].total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[4].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[5].passed", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[5].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[5].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[6].passed", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[6].total", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[6].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[7].passed", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[7].total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[7].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[8].passed", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[8].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[8].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[9].passed", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[9].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[9].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[10].passed", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[10].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[10].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[11].passed", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[11].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[11].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[12].passed", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[12].total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[12].skipped", "statistic": "single", "unit": "count", "value": 4 } ], "model": { "artifact_sha256": [], "publisher_repo": "ornith-ai/Ornith-1.5-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "structured-tasks", "observed": "71/130" }, { "met": false, "name": "API-cases", "observed": "71/124" } ], "reasons": [ "71/130 tasks passed.", "71/124 API cases passed." ], "status": "partial" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 71, "skipped": null, "suite": "structured-tasks", "total": 130, "version": "core-v1" }, { "passed": 71, "skipped": 4, "suite": "API-cases", "total": 124, "version": "core-v1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "42c4dc34a066f791a62c9b1dbeb97943988c746309a43a464f0417a02ae9f1da", "harness": { "name": "keep-api-suite-run-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": "core-v1", "workload_version": "1" }, "requests": [], "scenario": { "id": "api-suite", "version": "1" }, "settings": { "concurrency": 1, "context_tokens": 262144, "kv_dtype": "fp8", "speculative_decoding": "plain" }, "setup_complete": false, "setup_id": "14e1c30f3eeece6ca79232fac0816ef7", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp CUDA", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "73fcdc4200cc4a94ad6c3be0", "measured_at": "2026-10-07T01:26:57.117463+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 602.1789470130461 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 356.61622351699043 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 595.5385198440054 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 118.40441259101499 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 486.22535631700885 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.25827784312023766 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 96 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[0].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 118.39787794597214 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 238.27134855801705 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 597.0246858799946 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.29546763464632536 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 107 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[1].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 238.25885873904917 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 356.61622351699043 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 599.5231784170028 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.4487314908083534 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 110 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[2].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 356.60719916498056 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 477.851054742001 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 601.1813899719855 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8756969629459231 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[3].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 477.8489240729832 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 595.5385198440054 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 602.1688145390362 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.43969159948358 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 110 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[4].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 595.5340977389715 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 1.0893745169626852 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.8834583185593738 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 35.27370834350586 } ], "model": { "artifact_sha256": [], "publisher_repo": "nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16", "publisher_revision": "a9904d24bcc1d289a1950fa9d2b978c47cf903b9", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 595.5385198440054 } ], "reasons": [ "Observed decode overlap was 4 of 5 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "04e5839fffd3856049c11c680e5ae05a807f32bbf3dec72f7a00578188171f70", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.25827784312023766, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 118.39787794597214, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 96, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": true, "total_s": 486.22535631700885, "ttft_s": 118.40441259101499, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.29546763464632536, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 238.25885873904917, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 107, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": true, "total_s": 597.0246858799946, "ttft_s": 238.27134855801705, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.4487314908083534, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 356.60719916498056, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 110, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": true, "total_s": 599.5231784170028, "ttft_s": 356.61622351699043, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.8756969629459231, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 477.8489240729832, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": true, "total_s": 601.1813899719855, "ttft_s": 477.851054742001, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 16.43969159948358, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 595.5340977389715, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 110, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": true, "total_s": 602.1688145390362, "ttft_s": 595.5385198440054, "user_number": 0, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "warm", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "shared-record-groups-v2", "kv_dtype": "q8_0", "max_output_tokens": 512, "scenario": "shared", "thinking": false }, "setup_complete": false, "setup_id": "9e5118825b2267598d5691416d3ab69e", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": 250000, "document": "shared", "input_tokens": { "max": 250016, "median": 250016, "min": 250016 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "78fe8c2f559e476e974b3154", "measured_at": "2026-10-07T18:42:18.992576+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 4.077593544963747 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20195 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 3.9541576470364816 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.94106332701848 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20339 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 4.045734722982161 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.87543728000792 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20341 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 4.064152015023865 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.47356341474793 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20343 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 4.067299807036761 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.80042675925883 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[3].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20339 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[4].ttft_s", "statistic": "single", "unit": "s", "value": 4.055949887959287 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.79423518349562 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[4].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[4].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20339 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[5].ttft_s", "statistic": "single", "unit": "s", "value": 4.053401597018819 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.06771796964834 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[5].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[5].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20341 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[6].ttft_s", "statistic": "single", "unit": "s", "value": 4.067534626985434 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.32967759554883 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[6].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[6].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20340 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[7].ttft_s", "statistic": "single", "unit": "s", "value": 4.061455496004783 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.23758667883683 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[7].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[7].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20341 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[8].ttft_s", "statistic": "single", "unit": "s", "value": 4.077593544963747 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.96795350455105 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[8].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[8].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20346 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[9].ttft_s", "statistic": "single", "unit": "s", "value": 4.072965578001458 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.21962695135959 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[9].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[9].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20201 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[10].ttft_s", "statistic": "single", "unit": "s", "value": 3.951721183024347 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.84176392650568 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[10].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[10].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20342 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[11].ttft_s", "statistic": "single", "unit": "s", "value": 4.038304175948724 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.78939123540195 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[11].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[11].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20342 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[12].ttft_s", "statistic": "single", "unit": "s", "value": 4.044558615016285 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.98579988307088 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[12].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[12].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20343 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[13].ttft_s", "statistic": "single", "unit": "s", "value": 4.057327327958774 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.07657581560343 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[13].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[13].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20342 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[14].ttft_s", "statistic": "single", "unit": "s", "value": 4.058043524972163 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.69030762200015 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[14].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[14].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20343 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[15].ttft_s", "statistic": "single", "unit": "s", "value": 4.055920558981597 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.77333283588911 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[15].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[15].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20346 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[16].ttft_s", "statistic": "single", "unit": "s", "value": 4.056567955005448 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.97233699057458 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140010 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[16].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[16].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20343 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[17].ttft_s", "statistic": "single", "unit": "s", "value": 4.055032589996699 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.91166552338744 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[17].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[17].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20345 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[18].ttft_s", "statistic": "single", "unit": "s", "value": 4.065946794056799 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.04002594362656 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[18].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[18].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20344 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[19].ttft_s", "statistic": "single", "unit": "s", "value": 4.063116103003267 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.94838912770027 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[19].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[19].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 4.0576854264654685 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 3.9541576470364816 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 3.9529394150304142 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 20 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "18/20" } ], "reasons": [ "18/20 requests passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 2, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "47a03fd9bbfe612c42f992e302c0e2fe", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "7cde518077a3402d96c45d36", "measured_at": "2026-10-07T19:26:02.104296+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 34.95379966800101 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 49 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20208 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 3.9712129940162413 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.09795202635519 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40330 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 5.734351448016241 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.67000466206336 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4677411356310439 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60454 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 6.998552023957018 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 63.86763815721237 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6587488007410593 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80583 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 8.786693687026855 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 59.64226017200865 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[3].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7282925679113461 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100707 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[4].ttft_s", "statistic": "single", "unit": "s", "value": 10.51181011798326 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 56.93400500156006 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[4].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[4].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7908884188785288 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120833 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[5].ttft_s", "statistic": "single", "unit": "s", "value": 11.052641157992184 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 54.85910724382249 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[5].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[5].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8326202279178702 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80753 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[6].ttft_s", "statistic": "single", "unit": "s", "value": 27.630642823991366 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.650431781516401 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[6].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[6].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.13350223989922222 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100875 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[7].ttft_s", "statistic": "single", "unit": "s", "value": 10.670778055035044 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 46.23547286111225 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[7].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[7].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7895712515489467 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 121001 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[8].ttft_s", "statistic": "single", "unit": "s", "value": 16.25697027501883 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 40.93738964990943 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[8].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[8].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7739091139547444 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80759 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[9].ttft_s", "statistic": "single", "unit": "s", "value": 32.21716991101857 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.615524284125332 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[9].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[9].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5278871691546845 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20214 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[10].ttft_s", "statistic": "single", "unit": "s", "value": 8.937347330967896 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.07008524878014 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[10].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[10].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40334 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[11].ttft_s", "statistic": "single", "unit": "s", "value": 6.663203734031413 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.31468497382182 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[11].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[11].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6976325871580423 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60457 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[12].ttft_s", "statistic": "single", "unit": "s", "value": 15.628533505019732 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 27.0289973328533 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[12].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[12].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7739091139547444 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80585 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[13].ttft_s", "statistic": "single", "unit": "s", "value": 11.845644221000839 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.6756326677129403 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[13].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[13].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3637445458151527 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100712 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[14].ttft_s", "statistic": "single", "unit": "s", "value": 10.508183865982573 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 57.95413120272065 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[14].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[14].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7908491540233538 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120832 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[15].ttft_s", "statistic": "single", "unit": "s", "value": 11.055763568961993 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 54.77272820399646 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[15].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[15].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8326271186440678 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80760 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[16].ttft_s", "statistic": "single", "unit": "s", "value": 27.871361639001407 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.219512829778672 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140010 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[16].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[16].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.13349940553699824 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100884 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[17].ttft_s", "statistic": "single", "unit": "s", "value": 10.597855208034161 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.71815787264175 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[17].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[17].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7404662001772577 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 121014 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[18].ttft_s", "statistic": "single", "unit": "s", "value": 17.187022591999266 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 23.273090312646612 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[18].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[18].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7901626496163175 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80765 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[19].ttft_s", "statistic": "single", "unit": "s", "value": 33.59482376201777 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 19.805578457001552 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[19].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[19].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.596244364396548 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[20].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[20].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20209 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[20].ttft_s", "statistic": "single", "unit": "s", "value": 8.312428326986264 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[20].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.505895592028528 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[20].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[20].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[20].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[21].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[21].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40335 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[21].ttft_s", "statistic": "single", "unit": "s", "value": 6.018421126995236 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[21].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.134066828610283 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[21].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[21].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[21].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6975831863984308 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[22].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[22].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60463 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[22].ttft_s", "statistic": "single", "unit": "s", "value": 7.343568017007783 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[22].decode_tps", "statistic": "single", "unit": "tok/s", "value": 54.54647180882014 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[22].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[22].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[22].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6586507450837702 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[23].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[23].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80585 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[23].ttft_s", "statistic": "single", "unit": "s", "value": 17.834073767997324 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[23].decode_tps", "statistic": "single", "unit": "tok/s", "value": 38.673793931428015 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[23].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[23].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[23].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7901626496163175 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[24].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[24].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100714 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[24].ttft_s", "statistic": "single", "unit": "s", "value": 13.31568725197576 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[24].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.8041900838857634 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[24].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[24].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[24].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.43888273574352954 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[25].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[25].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120840 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[25].ttft_s", "statistic": "single", "unit": "s", "value": 11.462938445038162 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[25].decode_tps", "statistic": "single", "unit": "tok/s", "value": 54.78150041594344 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[25].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[25].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[25].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8325719960278054 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[26].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[26].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80763 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[26].ttft_s", "statistic": "single", "unit": "s", "value": 24.55534920503851 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[26].decode_tps", "statistic": "single", "unit": "tok/s", "value": 60.09149877948051 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[26].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[26].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[26].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[27].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[27].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100887 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[27].ttft_s", "statistic": "single", "unit": "s", "value": 10.169303657021374 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[27].decode_tps", "statistic": "single", "unit": "tok/s", "value": 57.11645550094017 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[27].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[27].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[27].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.789477336029419 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[28].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[28].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 121013 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[28].ttft_s", "statistic": "single", "unit": "s", "value": 11.176374671980739 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[28].decode_tps", "statistic": "single", "unit": "tok/s", "value": 54.48105135473672 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[28].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[28].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[28].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8313817523737119 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[29].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[29].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80758 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[29].ttft_s", "statistic": "single", "unit": "s", "value": 27.707559100992512 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[29].decode_tps", "statistic": "single", "unit": "tok/s", "value": 48.672904722500526 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[29].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[29].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[29].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[30].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[30].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20212 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[30].ttft_s", "statistic": "single", "unit": "s", "value": 10.21521253301762 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[30].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.395998915845764 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[30].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[30].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[30].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[31].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[31].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40337 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[31].ttft_s", "statistic": "single", "unit": "s", "value": 5.814369395026006 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[31].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.57164909704103 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[31].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[31].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[31].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4676599647965887 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[32].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[32].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60470 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[32].ttft_s", "statistic": "single", "unit": "s", "value": 7.021634981967509 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[32].decode_tps", "statistic": "single", "unit": "tok/s", "value": 63.87285075291422 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[32].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[32].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[32].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6585744997519432 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[33].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[33].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80594 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[33].ttft_s", "statistic": "single", "unit": "s", "value": 8.845477171009406 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[33].decode_tps", "statistic": "single", "unit": "tok/s", "value": 60.7653887621596 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[33].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[33].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[33].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7281931657443482 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[34].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[34].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100719 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[34].ttft_s", "statistic": "single", "unit": "s", "value": 10.101286827994045 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[34].decode_tps", "statistic": "single", "unit": "tok/s", "value": 57.33389323325864 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[34].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[34].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[34].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7907941897755141 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[35].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[35].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120840 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[35].ttft_s", "statistic": "single", "unit": "s", "value": 12.661408385029063 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[35].decode_tps", "statistic": "single", "unit": "tok/s", "value": 23.336457877781104 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[35].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[35].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[35].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7132587519673317 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[36].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[36].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80755 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[36].ttft_s", "statistic": "single", "unit": "s", "value": 29.363055747991893 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[36].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.913467639450548 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[36].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[36].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[36].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.32324834624938725 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[37].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[37].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100888 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[37].ttft_s", "statistic": "single", "unit": "s", "value": 10.620319917972665 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[37].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.802391710820769 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[37].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[37].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[37].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7622742150563705 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[38].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[38].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 121013 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[38].ttft_s", "statistic": "single", "unit": "s", "value": 18.540593829995487 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[38].decode_tps", "statistic": "single", "unit": "tok/s", "value": 23.95229356463117 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[38].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[38].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[38].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8129563566335481 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[39].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[39].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80770 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[39].ttft_s", "statistic": "single", "unit": "s", "value": 34.95379966800101 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[39].decode_tps", "statistic": "single", "unit": "tok/s", "value": 50.771285671428195 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[39].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[39].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[39].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4990129653694684 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[40].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[40].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20214 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[40].ttft_s", "statistic": "single", "unit": "s", "value": 10.346459145017434 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[40].decode_tps", "statistic": "single", "unit": "tok/s", "value": 43.404094618709124 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[40].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[40].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[40].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[41].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[41].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40338 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[41].ttft_s", "statistic": "single", "unit": "s", "value": 9.022375561995432 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[41].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.7966445694359456 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[41].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[41].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[41].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.15578109387000075 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[42].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[42].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60464 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[42].ttft_s", "statistic": "single", "unit": "s", "value": 7.315880263980944 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[42].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.328740724811907 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[42].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[42].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[42].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7404432544994795 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[43].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[43].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80590 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[43].ttft_s", "statistic": "single", "unit": "s", "value": 9.243037085980177 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[43].decode_tps", "statistic": "single", "unit": "tok/s", "value": 49.66064312639921 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[43].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[43].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[43].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7282293088472516 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[44].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[44].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100716 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[44].ttft_s", "statistic": "single", "unit": "s", "value": 19.10711804201128 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[44].decode_tps", "statistic": "single", "unit": "tok/s", "value": 37.511861146603756 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[44].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[44].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[44].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8129563566335481 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[45].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[45].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120844 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[45].ttft_s", "statistic": "single", "unit": "s", "value": 14.16911771998275 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[45].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.961791578354682 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[45].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[45].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[45].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4990129653694684 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[46].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[46].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80765 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[46].ttft_s", "statistic": "single", "unit": "s", "value": 24.98578308400465 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[46].decode_tps", "statistic": "single", "unit": "tok/s", "value": 60.04352009788808 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[46].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[46].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[46].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[47].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[47].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100891 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[47].ttft_s", "statistic": "single", "unit": "s", "value": 10.18339444999583 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[47].decode_tps", "statistic": "single", "unit": "tok/s", "value": 56.85279505394254 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[47].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[47].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[47].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7894460358208364 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[48].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[48].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 121017 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[48].ttft_s", "statistic": "single", "unit": "s", "value": 11.176227814052254 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[48].decode_tps", "statistic": "single", "unit": "tok/s", "value": 54.28764587131654 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[48].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[48].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[48].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8313542725402217 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[49].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[49].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80766 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[49].ttft_s", "statistic": "single", "unit": "s", "value": 24.5173131210031 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[49].decode_tps", "statistic": "single", "unit": "tok/s", "value": 59.898277960708626 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[49].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[49].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[49].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 27.78946036999696 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 19.10711804201128 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 10.427321505500004 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 50 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "49/50" } ], "reasons": [ "49/50 requests passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "73482781e9529432ec3a3f91003332b8", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Finite lab cohort; configured 262144 context differs from measured inputs. Aggregate uses first-output to final-completion window. Whole-host memory sampled during speed only; missing counters remain unassessed. Broader coding quality and delivery acceptance unassessed." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "7fa388c66cc846fd8bed4b9a", "measured_at": "2026-10-07T04:37:19.350179+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 79.81211304664612 }, { "basis": null, "n": 8, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 54.980975570506416 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": 8, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 65.05511922802543 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "preemptions", "scope": "machine_total", "source_field": "preemptions", "statistic": "single", "unit": "count", "value": 0.0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "mtp_acceptance_rate", "scope": "machine_total", "source_field": "mtp_acceptance_rate", "statistic": "single", "unit": "ratio", "value": 0.8618331053351573 }, { "basis": null, "n": null, "name": "correctness_probes_passed", "scope": "machine_total", "source_field": "correctness_probes_passed", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 149.84376022177028 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 51.320530727035084 }, { "basis": null, "n": null, "name": "correctness_probes_total", "scope": "machine_total", "source_field": "correctness_probes_total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": 8, "name": "decode_tps", "scope": "per_person", "source_field": "per_request_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 24.112197205943012 }, { "basis": null, "n": null, "name": "forced_output_tokens", "scope": "machine_total", "source_field": "forced_output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 93.4132308959961 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 54.981445715995505 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[0].total_s", "statistic": "single", "unit": "s", "value": 79.12366025097435 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 21.16624385304957 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40101 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 52.53015207097633 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[1].total_s", "statistic": "single", "unit": "s", "value": 79.22427673795028 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 19.142789148363075 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 54.9799332160037 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[2].total_s", "statistic": "single", "unit": "s", "value": 79.34082178003155 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 20.97624635722692 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[2].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 65.05511922802543 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[3].total_s", "statistic": "single", "unit": "s", "value": 79.81169429101283 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 34.62863149605059 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40096 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 54.14935153600527 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[4].total_s", "statistic": "single", "unit": "s", "value": 79.07180817902554 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 20.503596708758227 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[5].ttft_s", "statistic": "single", "unit": "s", "value": 64.94253399403533 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[5].total_s", "statistic": "single", "unit": "s", "value": 79.79206638300093 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 34.41185800434458 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[5].user_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[5].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[5].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[5].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[6].ttft_s", "statistic": "single", "unit": "s", "value": 54.98134611500427 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[6].total_s", "statistic": "single", "unit": "s", "value": 79.17577939503826 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 21.120560836681936 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[6].user_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[6].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[6].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[6].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[7].ttft_s", "statistic": "single", "unit": "s", "value": 54.98060502600856 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[7].total_s", "statistic": "single", "unit": "s", "value": 79.37474802898942 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 20.94765124306921 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[7].user_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40100 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[7].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[7].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[7].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 8, "name": "decode_tps", "scope": "per_person", "source_field": "per_person_decode_median_tps", "statistic": "median", "unit": "tok/s", "value": 21.04840359695443 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "8/8" }, { "met": true, "name": "hidden-facts", "observed": "24/24" }, { "met": true, "name": "correctness-probes", "observed": "8/8" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 24, "skipped": null, "suite": "hidden-facts", "total": 24, "version": "1" }, { "passed": 8, "skipped": null, "suite": "correctness-probes", "total": 8, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "e2097959b1ca0142f2b93edc3ec411e474c0a6622800dd247af7a71bd9c39e83", "harness": { "name": "keep-team-load-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 21.16624385304957, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40101, "total_s": 79.12366025097435, "ttft_s": 54.981445715995505, "user_number": 0 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 19.142789148363075, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 79.22427673795028, "ttft_s": 52.53015207097633, "user_number": 1 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 20.97624635722692, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 79.34082178003155, "ttft_s": 54.9799332160037, "user_number": 2 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 34.62863149605059, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40096, "total_s": 79.81169429101283, "ttft_s": 65.05511922802543, "user_number": 3 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 20.503596708758227, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 79.07180817902554, "ttft_s": 54.14935153600527, "user_number": 4 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 34.41185800434458, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 79.79206638300093, "ttft_s": 64.94253399403533, "user_number": 5 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 21.120560836681936, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 79.17577939503826, "ttft_s": 54.98134611500427, "user_number": 6 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 20.94765124306921, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40100, "total_s": 79.37474802898942, "ttft_s": 54.98060502600856, "user_number": 7 } ], "scenario": { "id": "team-load", "version": "1" }, "settings": { "concurrency": 8, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "cf0bbc6439f3c0ed87c21751d34d5655", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": { "max": 40105, "median": 40102.5, "min": 40096 }, "output_tokens": 512, "repetitions": null, "users": 8, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Missing named-source control: unknown source then original source omitted after a roll. Lab retrieval and rolling-history evidence only. No summaries. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "7fb62d7607014cd9b3edc0df", "measured_at": "2026-10-07T18:53:40.828001+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 4.035255178983789 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20199 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 3.953893987054471 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.05045780126063 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20309 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 4.018932530016173 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.17275618713863 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20307 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 4.035255178983789 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.5554045085893 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60010 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 4.027093854499981 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 3.953893987054471 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 3.953893987054471 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 3 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "3/3" } ], "reasons": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "47a03fd9bbfe612c42f992e302c0e2fe", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Measured retrieval only, with no customer qualification threshold. Exact dense float32 array size excludes payloads and vector database overhead. One timing run; no reranking, ingestion, answer generation or GPU measurement.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Transformers CPU", "recipe": "plain", "version": "5.19.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "84a23ce31fcb4c4bbe7855a5", "measured_at": "2026-10-07T14:39:45.935762+00:00", "metrics": [ { "basis": null, "n": null, "name": "passages_per_second", "scope": "machine_total", "source_field": "passages_per_second", "statistic": "single", "unit": "passages/s", "value": 12.25921414124026 }, { "basis": null, "n": null, "name": "query_seconds", "scope": "per_person", "source_field": "query_seconds", "statistic": "single", "unit": "s", "value": 2.574559000087902 }, { "basis": null, "n": null, "name": "recall_at_5", "scope": "machine_total", "source_field": "recall_at_5", "statistic": "single", "unit": "ratio", "value": 1.0 }, { "basis": null, "n": null, "name": "mrr", "scope": "machine_total", "source_field": "mrr", "statistic": "single", "unit": "ratio", "value": 0.9716666666666666 }, { "basis": null, "n": null, "name": "questions", "scope": "machine_total", "source_field": "questions", "statistic": "single", "unit": "count", "value": 100 }, { "basis": null, "n": null, "name": "passage_seconds", "scope": "machine_total", "source_field": "passage_seconds", "statistic": "single", "unit": "s", "value": 32.62851887498982 }, { "basis": null, "n": null, "name": "vector_bytes", "scope": "machine_total", "source_field": "vector_bytes", "statistic": "single", "unit": "bytes", "value": 1228800 }, { "basis": null, "n": null, "name": "passages", "scope": "machine_total", "source_field": "passages", "statistic": "single", "unit": "count", "value": 400 }, { "basis": null, "n": null, "name": "index_file_bytes", "scope": "machine_total", "source_field": "index_file_bytes", "statistic": "single", "unit": "bytes", "value": 1228928 }, { "basis": null, "n": null, "name": "recall_at_10", "scope": "machine_total", "source_field": "recall_at_10", "statistic": "single", "unit": "ratio", "value": 1.0 }, { "basis": null, "n": null, "name": "truncated_passages", "scope": "machine_total", "source_field": "truncated_passages", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "google/embeddinggemma-2", "publisher_revision": "914f7f89142e33e77833254d9c9b90c3cef7303b", "variant": "FP32 variant", "weights_dtype": "FP32", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-embedding-eval-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "embeddings", "version": "1" }, "settings": { "context_tokens": 262144, "corpus_sha256": "d7fb630b25a7d5128c44491d1a8f21bde479711f79520f1ce95ea8bd8a0ae1f6", "dimensions": 768, "dtype": "float32" }, "setup_complete": false, "setup_id": "a692739c4306f929629ef6c4a775d882", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple Silicon laptop CPU, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "861767d1989c495e851fd421", "measured_at": "2026-10-07T22:28:13.809032+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 83.85727065399988 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 105 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20208 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 21.29803962400183 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.783693721909644 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 21.609247204032727 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.792573166226104 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 34.670300349011086 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.229158992830583 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 27.17643065500306 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.136899158073063 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[3].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[4].ttft_s", "statistic": "single", "unit": "s", "value": 36.000922381994314 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.9590444949077743 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[4].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[4].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20375 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[5].ttft_s", "statistic": "single", "unit": "s", "value": 38.83920868905261 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.019755662023208 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[5].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[5].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[6].ttft_s", "statistic": "single", "unit": "s", "value": 40.54266342299525 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.201370846560778 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140015 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[6].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[6].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[7].ttft_s", "statistic": "single", "unit": "s", "value": 61.33349825098412 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.3125965460501154 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160018 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[7].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[7].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[8].ttft_s", "statistic": "single", "unit": "s", "value": 49.72121684002923 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0728841316990536 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[8].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[8].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[9].ttft_s", "statistic": "single", "unit": "s", "value": 50.01166466297582 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9904832327301951 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[9].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[9].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[10].ttft_s", "statistic": "single", "unit": "s", "value": 51.15083277004305 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.100283761532138 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[10].dropped_messages", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[10].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[11].ttft_s", "statistic": "single", "unit": "s", "value": 66.67075316002592 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.4481824986763667 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[11].dropped_messages", "statistic": "single", "unit": "count", "value": 22 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[11].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[12].ttft_s", "statistic": "single", "unit": "s", "value": 70.46918230800657 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.5922501627493335 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[12].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[12].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20384 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[13].ttft_s", "statistic": "single", "unit": "s", "value": 79.8193016120349 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.9255969933981314 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[13].dropped_messages", "statistic": "single", "unit": "count", "value": 26 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[13].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[14].ttft_s", "statistic": "single", "unit": "s", "value": 67.96495248703286 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.4639823594326746 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[14].dropped_messages", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[14].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[15].ttft_s", "statistic": "single", "unit": "s", "value": 63.044458357966505 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.2455207580803553 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[15].dropped_messages", "statistic": "single", "unit": "count", "value": 30 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[15].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[16].ttft_s", "statistic": "single", "unit": "s", "value": 52.80766814999515 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.115362845640841 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[16].dropped_messages", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[16].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[17].ttft_s", "statistic": "single", "unit": "s", "value": 51.723604545986746 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0476698448889976 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[17].dropped_messages", "statistic": "single", "unit": "count", "value": 34 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[17].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[18].ttft_s", "statistic": "single", "unit": "s", "value": 57.43843070202274 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.1806591159243558 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[18].dropped_messages", "statistic": "single", "unit": "count", "value": 36 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[18].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[19].ttft_s", "statistic": "single", "unit": "s", "value": 58.210758910980076 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.127450413604592 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[19].dropped_messages", "statistic": "single", "unit": "count", "value": 38 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[19].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[20].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[20].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20378 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[20].ttft_s", "statistic": "single", "unit": "s", "value": 52.639913316001184 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[20].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0340104973595965 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[20].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[20].dropped_messages", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[20].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[21].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[21].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20215 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[21].ttft_s", "statistic": "single", "unit": "s", "value": 36.14523050299613 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[21].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.86374648579424 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[21].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[21].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[21].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[22].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[22].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[22].ttft_s", "statistic": "single", "unit": "s", "value": 28.929401359986514 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[22].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.068068855379714 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[22].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40005 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[22].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[22].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[23].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[23].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20384 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[23].ttft_s", "statistic": "single", "unit": "s", "value": 45.35748997400515 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[23].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.2954410145949096 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[23].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[23].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[23].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[24].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[24].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[24].ttft_s", "statistic": "single", "unit": "s", "value": 40.25480733101722 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[24].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.3544498616016605 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[24].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[24].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[24].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[25].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[25].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20378 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[25].ttft_s", "statistic": "single", "unit": "s", "value": 41.776898667973 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[25].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.3421857321888933 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[25].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[25].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[25].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[26].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[26].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20384 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[26].ttft_s", "statistic": "single", "unit": "s", "value": 61.35739882104099 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[26].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.330292156279314 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[26].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[26].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[26].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[27].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[27].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[27].ttft_s", "statistic": "single", "unit": "s", "value": 51.82940217596479 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[27].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.064639599430281 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[27].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140015 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[27].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[27].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[28].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[28].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[28].ttft_s", "statistic": "single", "unit": "s", "value": 48.16035067598568 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[28].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0449510863507636 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[28].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160019 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[28].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[28].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[29].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[29].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20384 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[29].ttft_s", "statistic": "single", "unit": "s", "value": 52.919312131009065 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[29].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.1397815934678104 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[29].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[29].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[29].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[30].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[30].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[30].ttft_s", "statistic": "single", "unit": "s", "value": 69.09640868502902 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[30].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.5148565740342002 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[30].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[30].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[30].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[31].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[31].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[31].ttft_s", "statistic": "single", "unit": "s", "value": 70.50478764198488 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[31].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.6769084675392483 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[31].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[31].dropped_messages", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[31].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[32].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[32].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20387 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[32].ttft_s", "statistic": "single", "unit": "s", "value": 83.85727065399988 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[32].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.5001628453519287 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[32].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[32].dropped_messages", "statistic": "single", "unit": "count", "value": 22 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[32].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[33].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[33].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[33].ttft_s", "statistic": "single", "unit": "s", "value": 65.88525169203058 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[33].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8573151511422684 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[33].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[33].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[33].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[34].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[34].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[34].ttft_s", "statistic": "single", "unit": "s", "value": 22.14965163799934 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[34].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7448845980711477 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[34].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[34].dropped_messages", "statistic": "single", "unit": "count", "value": 26 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[34].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[35].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[35].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[35].ttft_s", "statistic": "single", "unit": "s", "value": 25.950526922009885 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[35].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8546376299740918 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[35].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[35].dropped_messages", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[35].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[36].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[36].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[36].ttft_s", "statistic": "single", "unit": "s", "value": 32.494666195008904 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[36].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8979574087394616 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[36].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[36].dropped_messages", "statistic": "single", "unit": "count", "value": 30 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[36].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[37].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[37].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[37].ttft_s", "statistic": "single", "unit": "s", "value": 22.200090466998518 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[37].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7218283696132785 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[37].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[37].dropped_messages", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[37].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[38].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[38].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[38].ttft_s", "statistic": "single", "unit": "s", "value": 23.037825768988114 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[38].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7504716338981225 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[38].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360027 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[38].dropped_messages", "statistic": "single", "unit": "count", "value": 34 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[38].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[39].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[39].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20385 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[39].ttft_s", "statistic": "single", "unit": "s", "value": 30.95947255101055 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[39].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.1066772684034836 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[39].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[39].dropped_messages", "statistic": "single", "unit": "count", "value": 36 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[39].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[40].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[40].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[40].ttft_s", "statistic": "single", "unit": "s", "value": 42.501848874031566 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[40].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.4063147798815891 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[40].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[40].dropped_messages", "statistic": "single", "unit": "count", "value": 38 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[40].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[41].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[41].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20384 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[41].ttft_s", "statistic": "single", "unit": "s", "value": 42.54126005800208 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[41].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.2976804989944621 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[41].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420032 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[41].dropped_messages", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[41].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[42].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[42].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20213 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[42].ttft_s", "statistic": "single", "unit": "s", "value": 32.83461865299614 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[42].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.6226829330423767 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[42].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[42].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[42].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[43].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[43].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[43].ttft_s", "statistic": "single", "unit": "s", "value": 21.976252384018153 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[43].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.3643714364808432 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[43].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[43].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[43].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[44].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[44].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20378 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[44].ttft_s", "statistic": "single", "unit": "s", "value": 21.99750637699617 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[44].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9078830712083121 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[44].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[44].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[44].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[45].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[45].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[45].ttft_s", "statistic": "single", "unit": "s", "value": 42.37184034602251 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[45].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9783759598966434 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[45].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[45].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[45].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[46].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[46].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[46].ttft_s", "statistic": "single", "unit": "s", "value": 25.117952532018535 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[46].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8397508956720273 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[46].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[46].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[46].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[47].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[47].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[47].ttft_s", "statistic": "single", "unit": "s", "value": 34.90033710398711 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[47].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8064433328764812 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[47].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120017 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[47].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[47].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[48].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[48].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[48].ttft_s", "statistic": "single", "unit": "s", "value": 29.66354861902073 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[48].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9837727761142137 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[48].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140019 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[48].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[48].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[49].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[49].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[49].ttft_s", "statistic": "single", "unit": "s", "value": 41.8535802050028 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[49].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.019066252153321 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[49].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160022 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[49].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[49].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[50].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[50].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[50].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[50].ttft_s", "statistic": "single", "unit": "s", "value": 46.80564466700889 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[50].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.11869815866569 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[50].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[50].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[50].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[50].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[51].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[51].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[51].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20378 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[51].ttft_s", "statistic": "single", "unit": "s", "value": 79.61618936801096 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[51].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.3555220087627093 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[51].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[51].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[51].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[51].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[52].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[52].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[52].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[52].ttft_s", "statistic": "single", "unit": "s", "value": 71.31167039799038 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[52].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.5820780983691458 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[52].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[52].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[52].dropped_messages", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[52].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[53].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[53].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[53].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[53].ttft_s", "statistic": "single", "unit": "s", "value": 63.075886620034 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[53].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.282322391011998 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[53].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[53].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[53].dropped_messages", "statistic": "single", "unit": "count", "value": 22 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[53].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[54].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[54].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[54].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[54].ttft_s", "statistic": "single", "unit": "s", "value": 54.74592198303435 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[54].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.125289550384349 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[54].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[54].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260027 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[54].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[54].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[55].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[55].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[55].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[55].ttft_s", "statistic": "single", "unit": "s", "value": 51.76248636102537 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[55].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0933835856318905 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[55].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[55].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[55].dropped_messages", "statistic": "single", "unit": "count", "value": 26 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[55].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[56].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[56].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[56].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[56].ttft_s", "statistic": "single", "unit": "s", "value": 59.44338772603078 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[56].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.2149839896544588 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[56].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[56].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[56].dropped_messages", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[56].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[57].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[57].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[57].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[57].ttft_s", "statistic": "single", "unit": "s", "value": 55.94809865200659 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[57].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.163241641611616 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[57].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[57].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320030 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[57].dropped_messages", "statistic": "single", "unit": "count", "value": 30 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[57].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[58].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[58].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[58].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[58].ttft_s", "statistic": "single", "unit": "s", "value": 54.55966194003122 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[58].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.18672476847471 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[58].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[58].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340030 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[58].dropped_messages", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[58].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[59].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[59].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[59].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[59].ttft_s", "statistic": "single", "unit": "s", "value": 42.43145718204323 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[59].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.3326528665371615 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[59].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[59].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360030 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[59].dropped_messages", "statistic": "single", "unit": "count", "value": 34 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[59].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[60].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[60].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[60].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[60].ttft_s", "statistic": "single", "unit": "s", "value": 42.41143779596314 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[60].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.368446110791211 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[60].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[60].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380031 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[60].dropped_messages", "statistic": "single", "unit": "count", "value": 36 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[60].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[61].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[61].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[61].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20377 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[61].ttft_s", "statistic": "single", "unit": "s", "value": 42.5212757140398 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[61].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.237045252696843 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[61].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[61].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400031 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[61].dropped_messages", "statistic": "single", "unit": "count", "value": 38 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[61].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[62].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[62].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[62].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[62].ttft_s", "statistic": "single", "unit": "s", "value": 40.6386415809975 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[62].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.194488168197656 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[62].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[62].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420031 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[62].dropped_messages", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[62].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[63].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[63].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[63].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20214 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[63].ttft_s", "statistic": "single", "unit": "s", "value": 58.5853517089854 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[63].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.1035870734476796 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[63].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[63].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[63].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[63].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[64].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[64].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[64].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20385 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[64].ttft_s", "statistic": "single", "unit": "s", "value": 43.77859455102589 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[64].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.976996814962709 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[64].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[64].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[64].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[64].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[65].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[65].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[65].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[65].ttft_s", "statistic": "single", "unit": "s", "value": 46.71583229204407 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[65].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0957480235099495 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[65].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[65].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60005 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[65].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[65].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[66].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[66].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[66].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20378 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[66].ttft_s", "statistic": "single", "unit": "s", "value": 45.31801800901303 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[66].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.401064233535292 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[66].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[66].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80005 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[66].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[66].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[67].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[67].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[67].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[67].ttft_s", "statistic": "single", "unit": "s", "value": 68.71087204298237 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[67].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.4521721765276623 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[67].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[67].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[67].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[67].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[68].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[68].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[68].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20384 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[68].ttft_s", "statistic": "single", "unit": "s", "value": 68.2611089880229 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[68].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.5584375020447734 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[68].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[68].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[68].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[68].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[69].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[69].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[69].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[69].ttft_s", "statistic": "single", "unit": "s", "value": 82.24285914597567 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[69].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.8785416097644905 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[69].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[69].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[69].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[69].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[70].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[70].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[70].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20386 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[70].ttft_s", "statistic": "single", "unit": "s", "value": 75.42471403401578 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[70].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.2162122167507055 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[70].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[70].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[70].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[70].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[71].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[71].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[71].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20385 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[71].ttft_s", "statistic": "single", "unit": "s", "value": 63.46974950900767 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[71].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0889460659473411 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[71].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[71].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180016 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[71].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[71].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[72].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[72].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[72].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20387 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[72].ttft_s", "statistic": "single", "unit": "s", "value": 51.69210169499274 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[72].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0884265610810726 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[72].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[72].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200018 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[72].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[72].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[73].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[73].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[73].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[73].ttft_s", "statistic": "single", "unit": "s", "value": 51.461301664006896 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[73].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.3999652223427503 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[73].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[73].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220018 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[73].dropped_messages", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[73].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[74].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[74].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[74].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20385 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[74].ttft_s", "statistic": "single", "unit": "s", "value": 56.930191038001794 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[74].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.2237868100453688 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[74].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[74].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240018 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[74].dropped_messages", "statistic": "single", "unit": "count", "value": 22 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[74].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[75].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[75].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[75].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20386 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[75].ttft_s", "statistic": "single", "unit": "s", "value": 53.95164775499143 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[75].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0468512810076605 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[75].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[75].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[75].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[75].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[76].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[76].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[76].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[76].ttft_s", "statistic": "single", "unit": "s", "value": 47.534067681001034 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[76].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.4107799423327996 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[76].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[76].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[76].dropped_messages", "statistic": "single", "unit": "count", "value": 26 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[76].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[77].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[77].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[77].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20384 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[77].ttft_s", "statistic": "single", "unit": "s", "value": 42.73077383998316 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[77].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.4263403916990791 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[77].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[77].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[77].dropped_messages", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[77].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[78].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[78].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[78].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[78].ttft_s", "statistic": "single", "unit": "s", "value": 42.762363703979645 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[78].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.3883327917796255 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[78].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[78].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[78].dropped_messages", "statistic": "single", "unit": "count", "value": 30 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[78].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[79].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[79].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[79].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[79].ttft_s", "statistic": "single", "unit": "s", "value": 43.261375803966075 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[79].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.4337881560959045 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[79].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[79].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[79].dropped_messages", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[79].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[80].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[80].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[80].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20384 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[80].ttft_s", "statistic": "single", "unit": "s", "value": 41.75088982703164 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[80].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.5794254299048114 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[80].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[80].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[80].dropped_messages", "statistic": "single", "unit": "count", "value": 34 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[80].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[81].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[81].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[81].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[81].ttft_s", "statistic": "single", "unit": "s", "value": 22.06368842703523 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[81].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.5177788832525887 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[81].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[81].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[81].dropped_messages", "statistic": "single", "unit": "count", "value": 36 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[81].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[82].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[82].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[82].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[82].ttft_s", "statistic": "single", "unit": "s", "value": 21.987059842969757 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[82].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.520577556200404 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[82].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[82].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400026 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[82].dropped_messages", "statistic": "single", "unit": "count", "value": 38 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[82].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[83].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[83].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[83].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20387 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[83].ttft_s", "statistic": "single", "unit": "s", "value": 22.086384548980277 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[83].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.6388517235608977 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[83].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[83].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420027 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[83].dropped_messages", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[83].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[84].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[84].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[84].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20213 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[84].ttft_s", "statistic": "single", "unit": "s", "value": 42.39955521299271 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[84].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8795050630688763 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[84].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[84].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[84].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[84].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[85].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[85].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[85].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[85].ttft_s", "statistic": "single", "unit": "s", "value": 24.213497836026363 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[85].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7545661486089743 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[85].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[85].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[85].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[85].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[86].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[86].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[86].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[86].ttft_s", "statistic": "single", "unit": "s", "value": 25.909822629007977 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[86].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8090588520810669 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[86].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[86].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[86].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[86].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[87].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[87].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[87].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20386 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[87].ttft_s", "statistic": "single", "unit": "s", "value": 29.324067985988222 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[87].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7434827372262703 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[87].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[87].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[87].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[87].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[88].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[88].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[88].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[88].ttft_s", "statistic": "single", "unit": "s", "value": 22.145986246003304 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[88].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7528398150678814 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[88].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[88].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[88].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[88].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[89].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[89].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[89].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20385 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[89].ttft_s", "statistic": "single", "unit": "s", "value": 22.25557405903237 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[89].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7829850155172352 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[89].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[89].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[89].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[89].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[90].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[90].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[90].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[90].ttft_s", "statistic": "single", "unit": "s", "value": 22.179980229004286 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[90].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7199470312811947 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[90].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[90].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[90].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[90].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[91].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[91].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[91].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20386 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[91].ttft_s", "statistic": "single", "unit": "s", "value": 22.288278679014184 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[91].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7639840564285135 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[91].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[91].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160017 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[91].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[91].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[92].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[92].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[92].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[92].ttft_s", "statistic": "single", "unit": "s", "value": 22.218704838014673 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[92].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7420516384967465 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[92].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[92].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[92].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[92].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[93].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[93].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[93].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[93].ttft_s", "statistic": "single", "unit": "s", "value": 22.151969764963724 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[93].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.741497661062032 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[93].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[93].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[93].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[93].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[94].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[94].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[94].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[94].ttft_s", "statistic": "single", "unit": "s", "value": 22.155064663966186 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[94].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7325611374587515 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[94].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[94].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[94].dropped_messages", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[94].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[95].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[95].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[95].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[95].ttft_s", "statistic": "single", "unit": "s", "value": 22.222309100965504 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[95].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7205297254312101 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[95].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[95].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[95].dropped_messages", "statistic": "single", "unit": "count", "value": 22 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[95].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[96].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[96].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[96].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[96].ttft_s", "statistic": "single", "unit": "s", "value": 24.166431215999182 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[96].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.837786468631022 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[96].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[96].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[96].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[96].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[97].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[97].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[97].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[97].ttft_s", "statistic": "single", "unit": "s", "value": 30.919274711981416 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[97].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.352788298277453 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[97].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[97].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[97].dropped_messages", "statistic": "single", "unit": "count", "value": 26 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[97].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[98].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[98].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[98].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20385 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[98].ttft_s", "statistic": "single", "unit": "s", "value": 42.889956241007894 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[98].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.4269509820171984 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[98].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[98].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300024 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[98].dropped_messages", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[98].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[99].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[99].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[99].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20388 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[99].ttft_s", "statistic": "single", "unit": "s", "value": 43.08174711599713 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[99].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.3551379089992681 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[99].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[99].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320028 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[99].dropped_messages", "statistic": "single", "unit": "count", "value": 30 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[99].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[100].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[100].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[100].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20387 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[100].ttft_s", "statistic": "single", "unit": "s", "value": 22.212329739995766 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[100].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.364010998131154 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[100].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[100].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340031 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[100].dropped_messages", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[100].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[101].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[101].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[101].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20386 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[101].ttft_s", "statistic": "single", "unit": "s", "value": 22.206508811970707 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[101].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.5991180325795655 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[101].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[101].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[101].dropped_messages", "statistic": "single", "unit": "count", "value": 34 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[101].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[102].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[102].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[102].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20384 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[102].ttft_s", "statistic": "single", "unit": "s", "value": 22.024152108991984 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[102].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.548538786642081 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[102].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[102].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380034 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[102].dropped_messages", "statistic": "single", "unit": "count", "value": 36 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[102].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[103].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[103].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[103].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[103].ttft_s", "statistic": "single", "unit": "s", "value": 21.93195597396698 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[103].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.550793438712979 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[103].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[103].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400034 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[103].dropped_messages", "statistic": "single", "unit": "count", "value": 38 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[103].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[104].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[104].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[104].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[104].ttft_s", "statistic": "single", "unit": "s", "value": 22.018456218007486 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[104].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.644756333717181 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[104].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[104].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420034 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[104].dropped_messages", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[104].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 42.63601694899262 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 58.5853517089854 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 100 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 36.14523050299613 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 100 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 105 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "105/105" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "9fbdb045015b3a6790f7fe36b7e05c70", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Startup, inference completion, concurrent decode and retrieval correctness are separate gates.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "88cc6b909da044e3af0144d5", "measured_at": "2026-10-07T04:31:41.690098+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/0" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "did_not_start" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "long-context-concurrency", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "a9b2513c7b4da2c5c126534490dcbefe", "standard": "pre-standard", "system": { "driver": null, "hardware": "unrecorded hardware", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Startup, inference completion, concurrent decode and retrieval correctness are separate gates.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "893f89d85d0a42efaa1f8a11", "measured_at": "2026-10-07T08:48:23.100755+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/0" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "did_not_start" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "long-context-concurrency", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "2cdb4314c031e68844e1f2572dc8372d", "standard": "pre-standard", "system": { "driver": null, "hardware": "unrecorded hardware", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "8a22159d4ed2420f8fe7735f", "measured_at": "2026-10-07T21:47:51.419664+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 275.17675199900987 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 50 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20208 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 21.34866668097675 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.774761614646575 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40334 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 25.998813118960243 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.306212889593137 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4665046858729608 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60462 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 48.94798382301815 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.604281349957903 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4858941940601914 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80595 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 56.13542206399143 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.814570004771749 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[3].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6482928563750175 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100723 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[4].ttft_s", "statistic": "single", "unit": "s", "value": 82.31287157000042 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.766773324205591 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[4].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[4].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.656945814980402 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120853 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[5].ttft_s", "statistic": "single", "unit": "s", "value": 91.19235513999593 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.463936041009317 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[5].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[5].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6822569379506185 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80765 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[6].ttft_s", "statistic": "single", "unit": "s", "value": 212.87253164499998 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.5961379072053026 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[6].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[6].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5885301957049413 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100891 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[7].ttft_s", "statistic": "single", "unit": "s", "value": 36.43620048899902 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.5857785088552879 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[7].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[7].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6109509269820129 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 121021 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[8].ttft_s", "statistic": "single", "unit": "s", "value": 70.17824865295552 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7207588806955607 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[8].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[8].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8039733659721467 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80771 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[9].ttft_s", "statistic": "single", "unit": "s", "value": 174.00996459898306 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.39872927020492926 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[9].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[9].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.596028241267155 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20214 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[10].ttft_s", "statistic": "single", "unit": "s", "value": 48.61080889898585 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.603009872980028 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[10].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[10].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4858941940601914 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40338 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[11].ttft_s", "statistic": "single", "unit": "s", "value": 55.963811819965485 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.137415677786839 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[11].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[11].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6482928563750175 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60465 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[12].ttft_s", "statistic": "single", "unit": "s", "value": 82.19517153897323 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.462361372281482 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[12].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[12].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.656945814980402 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80597 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[13].ttft_s", "statistic": "single", "unit": "s", "value": 93.22804822097532 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.7146014041743376 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[13].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[13].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6822569379506185 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100728 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[14].ttft_s", "statistic": "single", "unit": "s", "value": 122.79134114802582 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7917546302519759 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[14].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[14].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.45611680588657244 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120852 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[15].ttft_s", "statistic": "single", "unit": "s", "value": 118.00764820800396 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.6934002427321664 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[15].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[15].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7389621986308166 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80772 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[16].ttft_s", "statistic": "single", "unit": "s", "value": 235.59306602203287 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.6041135674661237 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140010 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[16].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[16].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6279064421724018 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100900 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[17].ttft_s", "statistic": "single", "unit": "s", "value": 65.22222149197478 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.48597981714284966 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[17].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[17].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6454483010015697 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 121034 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[18].ttft_s", "statistic": "single", "unit": "s", "value": 70.11118865699973 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.5434399946335633 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[18].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[18].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6220197418375095 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80777 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[19].ttft_s", "statistic": "single", "unit": "s", "value": 269.8233214970096 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.02537681151602 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[19].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[19].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4702078281050353 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[20].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[20].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20209 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[20].ttft_s", "statistic": "single", "unit": "s", "value": 81.09656125702895 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[20].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.450961935024761 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[20].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[20].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[20].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.656945814980402 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[21].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[21].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40339 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[21].ttft_s", "statistic": "single", "unit": "s", "value": 93.25102354696719 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[21].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.7406196829265177 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[21].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[21].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[21].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6822569379506185 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[22].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[22].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60471 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[22].ttft_s", "statistic": "single", "unit": "s", "value": 127.11410134704784 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[22].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8395794969639764 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[22].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[22].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[22].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.45611680588657244 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[23].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[23].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80597 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[23].ttft_s", "statistic": "single", "unit": "s", "value": 119.55422945698956 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[23].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.7607409426385143 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[23].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[23].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[23].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7389621986308166 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[24].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[24].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100730 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[24].ttft_s", "statistic": "single", "unit": "s", "value": 115.67428891296731 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[24].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.6289571900990063 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[24].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[24].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[24].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6291780225923747 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[25].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[25].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120860 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[25].ttft_s", "statistic": "single", "unit": "s", "value": 109.1750186390127 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[25].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.5675222556055967 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[25].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[25].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[25].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7090295693589999 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[26].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[26].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80775 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[26].ttft_s", "statistic": "single", "unit": "s", "value": 206.19518313999288 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[26].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8880047422937126 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[26].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[26].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[26].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4144918497285308 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[27].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[27].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100903 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[27].ttft_s", "statistic": "single", "unit": "s", "value": 94.3664840660058 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[27].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7163421468716464 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[27].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[27].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[27].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7925235126804951 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[28].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[28].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 121033 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[28].ttft_s", "statistic": "single", "unit": "s", "value": 63.804067397955805 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[28].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.6746434211214533 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[28].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[28].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[28].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.595677768454713 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[29].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[29].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80770 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[29].ttft_s", "statistic": "single", "unit": "s", "value": 177.91871266497765 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[29].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.2128792267597928 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[29].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[29].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[29].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5563049894059608 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[30].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[30].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20212 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[30].ttft_s", "statistic": "single", "unit": "s", "value": 23.81672389700543 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[30].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.166082347920344 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[30].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[30].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[30].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[31].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[31].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40341 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[31].ttft_s", "statistic": "single", "unit": "s", "value": 126.18962677498348 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[31].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9055876004601872 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[31].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[31].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[31].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5629483360929303 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[32].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[32].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60478 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[32].ttft_s", "statistic": "single", "unit": "s", "value": 118.52998629101785 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[32].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.7473934670952544 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[32].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[32].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[32].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6998313898295478 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[33].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[33].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80606 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[33].ttft_s", "statistic": "single", "unit": "s", "value": 114.78437291301088 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[33].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.6289367728617169 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[33].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[33].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[33].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6291780225923747 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[34].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[34].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100735 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[34].ttft_s", "statistic": "single", "unit": "s", "value": 107.12015071400674 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[34].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.5303989401045802 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[34].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[34].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[34].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6610728545801747 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[35].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[35].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120860 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[35].ttft_s", "statistic": "single", "unit": "s", "value": 89.23787358897971 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[35].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.6250691616888681 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[35].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[35].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[35].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.662898765000555 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[36].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[36].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80767 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[36].ttft_s", "statistic": "single", "unit": "s", "value": 275.17675199900987 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[36].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9949640664567648 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[36].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[36].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[36].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.47968198030834647 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[37].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[37].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100904 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[37].ttft_s", "statistic": "single", "unit": "s", "value": 67.30511702201329 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[37].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.6838488729188765 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[37].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[37].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[37].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4401730572343869 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[38].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[38].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 121033 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[38].ttft_s", "statistic": "single", "unit": "s", "value": 64.33838980802102 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[38].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7130699661136162 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[38].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[38].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[38].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8124718392358296 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[39].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[39].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80782 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[39].ttft_s", "statistic": "single", "unit": "s", "value": 143.36928282101871 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[39].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.8319405845869736 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[39].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[39].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[39].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3551063882489906 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[40].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[40].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20214 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[40].ttft_s", "statistic": "single", "unit": "s", "value": 108.41189420601586 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[40].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9563096227186492 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[40].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[40].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[40].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[41].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[41].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40342 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[41].ttft_s", "statistic": "single", "unit": "s", "value": 119.34907640505116 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[41].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.7441345826868442 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[41].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[41].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[41].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7389621986308166 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[42].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[42].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60472 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[42].ttft_s", "statistic": "single", "unit": "s", "value": 113.27230199903715 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[42].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.6245213128105654 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[42].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[42].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[42].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6291780225923747 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[43].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[43].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80602 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[43].ttft_s", "statistic": "single", "unit": "s", "value": 109.211098874046 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[43].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.5784338295091597 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[43].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[43].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[43].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6610728545801747 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[44].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[44].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100732 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[44].ttft_s", "statistic": "single", "unit": "s", "value": 90.43948712199926 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[44].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.6346421028413529 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[44].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[44].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[44].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.662898765000555 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[45].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[45].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120864 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[45].ttft_s", "statistic": "single", "unit": "s", "value": 79.00042873196071 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[45].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.41281054151765945 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[45].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[45].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[45].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4704274075098288 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[46].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[46].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80777 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[46].ttft_s", "statistic": "single", "unit": "s", "value": 178.908187388035 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[46].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.0503554859383222 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[46].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[46].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[46].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5563182211897716 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[47].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[47].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100907 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[47].ttft_s", "statistic": "single", "unit": "s", "value": 66.3480821030098 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[47].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.6192819835386087 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[47].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[47].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[47].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4401367171375262 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[48].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[48].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 121037 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[48].ttft_s", "statistic": "single", "unit": "s", "value": 64.1394479930168 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[48].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.9848474347919035 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[48].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[48].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[48].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8291018448904054 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[49].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[49].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80778 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[49].ttft_s", "statistic": "single", "unit": "s", "value": 105.54750945296837 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[49].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.485465086732184 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[49].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[49].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[49].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 192.55168526401394 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 127.11410134704784 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 85.77537257949007 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 50 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "50/50" } ], "reasons": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "b1d6c72e429d1f589f72ab11c0605970", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "90c568b5a488418c98b60943", "measured_at": "2026-10-07T19:14:45.660805+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 15.419391048024409 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 49 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20208 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 3.9585845199762844 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.56483663007187 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40330 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 5.717325013014488 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.6685027704658 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4677411356310439 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60454 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 6.967021174961701 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 63.381646168488864 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6587488007410593 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40508 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 9.557411580986809 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.26706877832474 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[3].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60632 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[4].ttft_s", "statistic": "single", "unit": "s", "value": 6.999481018981896 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 63.63191239791189 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[4].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[4].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6568148832299776 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40500 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[5].ttft_s", "statistic": "single", "unit": "s", "value": 9.584961790998932 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 68.17714136663722 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[5].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[5].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60625 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[6].ttft_s", "statistic": "single", "unit": "s", "value": 7.004340023966506 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 63.89447167427353 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[6].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[6].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6568907216494846 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40501 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[7].ttft_s", "statistic": "single", "unit": "s", "value": 13.095244421972893 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 54.07236803025387 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[7].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[7].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60627 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[8].ttft_s", "statistic": "single", "unit": "s", "value": 11.723893465998117 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 48.00286443767311 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[8].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[8].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5812937669000902 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40511 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[9].ttft_s", "statistic": "single", "unit": "s", "value": 15.419391048024409 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 49.56860994869211 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[9].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[9].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3944219950875525 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20214 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[10].ttft_s", "statistic": "single", "unit": "s", "value": 8.198941089969594 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 15.516585144437444 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[10].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[10].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40334 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[11].ttft_s", "statistic": "single", "unit": "s", "value": 11.872520234028343 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 44.7314299641847 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[11].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[11].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5812937669000902 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60457 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[12].ttft_s", "statistic": "single", "unit": "s", "value": 9.798050606972538 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.452182121977078 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[12].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[12].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3944219950875525 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40506 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[13].ttft_s", "statistic": "single", "unit": "s", "value": 9.60545260400977 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.72896022677112 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[13].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[13].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60631 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[14].ttft_s", "statistic": "single", "unit": "s", "value": 7.041359800030477 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 64.19878826484305 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[14].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[14].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6568257162177764 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40505 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[15].ttft_s", "statistic": "single", "unit": "s", "value": 9.586103357025422 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.96500071903824 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[15].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[15].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60634 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[16].ttft_s", "statistic": "single", "unit": "s", "value": 7.040732189023402 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 63.75751287083469 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140010 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[16].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[16].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6567932183263515 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40508 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[17].ttft_s", "statistic": "single", "unit": "s", "value": 9.57341983600054 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.69124581849265 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[17].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[17].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60638 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[18].ttft_s", "statistic": "single", "unit": "s", "value": 7.024900424003135 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 63.981314385228444 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[18].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[18].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.656749892806491 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40510 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[19].ttft_s", "statistic": "single", "unit": "s", "value": 9.613097972993273 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.9359841242124 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[19].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[19].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[20].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[20].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20209 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[20].ttft_s", "statistic": "single", "unit": "s", "value": 3.987209509999957 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[20].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.7908042927028 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[20].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[20].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[20].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[21].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[21].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40335 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[21].ttft_s", "statistic": "single", "unit": "s", "value": 5.731002334970981 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[21].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.5149900934654 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[21].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[21].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[21].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.46768315358869467 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[22].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[22].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60463 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[22].ttft_s", "statistic": "single", "unit": "s", "value": 7.0075611409847625 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[22].decode_tps", "statistic": "single", "unit": "tok/s", "value": 63.96712841607582 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[22].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[22].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[22].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6586507450837702 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[23].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[23].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40503 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[23].ttft_s", "statistic": "single", "unit": "s", "value": 9.579465680988505 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[23].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.38209041030751 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[23].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[23].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[23].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[24].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[24].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60634 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[24].ttft_s", "statistic": "single", "unit": "s", "value": 7.02573576499708 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[24].decode_tps", "statistic": "single", "unit": "tok/s", "value": 63.9827153477398 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[24].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[24].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[24].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6567932183263515 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[25].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[25].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40509 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[25].ttft_s", "statistic": "single", "unit": "s", "value": 9.588259076001123 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[25].decode_tps", "statistic": "single", "unit": "tok/s", "value": 68.0236130924791 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[25].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[25].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[25].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[26].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[26].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60637 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[26].ttft_s", "statistic": "single", "unit": "s", "value": 7.020835597999394 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[26].decode_tps", "statistic": "single", "unit": "tok/s", "value": 63.71103897251925 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[26].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[26].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[26].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6567607236505764 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[27].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[27].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40503 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[27].ttft_s", "statistic": "single", "unit": "s", "value": 9.569037739012856 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[27].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.6141551191712 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[27].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[27].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[27].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[28].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[28].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60629 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[28].ttft_s", "statistic": "single", "unit": "s", "value": 7.043798096012324 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[28].decode_tps", "statistic": "single", "unit": "tok/s", "value": 64.34398681320698 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[28].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[28].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[28].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.656847383265434 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[29].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[29].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40509 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[29].ttft_s", "statistic": "single", "unit": "s", "value": 9.59583908499917 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[29].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.92341773706606 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[29].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[29].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[29].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[30].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[30].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20212 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[30].ttft_s", "statistic": "single", "unit": "s", "value": 3.9604154949774966 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[30].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.80392286852782 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[30].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[30].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[30].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[31].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[31].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40337 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[31].ttft_s", "statistic": "single", "unit": "s", "value": 5.718336247955449 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[31].decode_tps", "statistic": "single", "unit": "tok/s", "value": 68.0484381370113 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[31].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[31].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[31].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4676599647965887 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[32].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[32].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60470 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[32].ttft_s", "statistic": "single", "unit": "s", "value": 6.963763076986652 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[32].decode_tps", "statistic": "single", "unit": "tok/s", "value": 64.25111163565307 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[32].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[32].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[32].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6585744997519432 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[33].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[33].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40506 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[33].ttft_s", "statistic": "single", "unit": "s", "value": 9.553510278987233 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[33].decode_tps", "statistic": "single", "unit": "tok/s", "value": 68.13773324337446 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[33].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[33].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[33].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[34].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[34].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60629 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[34].ttft_s", "statistic": "single", "unit": "s", "value": 7.019993420981336 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[34].decode_tps", "statistic": "single", "unit": "tok/s", "value": 64.31336435803361 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[34].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[34].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[34].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.656847383265434 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[35].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[35].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40501 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[35].ttft_s", "statistic": "single", "unit": "s", "value": 9.54359990003286 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[35].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.65418541306987 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[35].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[35].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[35].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[36].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[36].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60629 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[36].ttft_s", "statistic": "single", "unit": "s", "value": 7.012918577995151 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[36].decode_tps", "statistic": "single", "unit": "tok/s", "value": 63.99978458218026 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[36].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[36].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[36].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.656847383265434 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[37].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[37].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40514 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[37].ttft_s", "statistic": "single", "unit": "s", "value": 12.776564086962026 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[37].decode_tps", "statistic": "single", "unit": "tok/s", "value": 53.21693503236824 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[37].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[37].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[37].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[38].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[38].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60639 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[38].ttft_s", "statistic": "single", "unit": "s", "value": 11.356965953949839 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[38].decode_tps", "statistic": "single", "unit": "tok/s", "value": 47.814449004601926 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[38].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[38].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[38].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5812016597839111 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[39].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[39].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40514 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[39].ttft_s", "statistic": "single", "unit": "s", "value": 15.259857495024335 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[39].decode_tps", "statistic": "single", "unit": "tok/s", "value": 50.99302923788085 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[39].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[39].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[39].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.39438293489671017 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[40].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[40].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20214 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[40].ttft_s", "statistic": "single", "unit": "s", "value": 7.772708332980983 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[40].decode_tps", "statistic": "single", "unit": "tok/s", "value": 15.279374927179404 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[40].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[40].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[40].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[41].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[41].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40338 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[41].ttft_s", "statistic": "single", "unit": "s", "value": 11.43570426700171 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[41].decode_tps", "statistic": "single", "unit": "tok/s", "value": 44.106299791193706 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[41].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[41].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[41].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5812016597839111 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[42].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[42].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60464 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[42].ttft_s", "statistic": "single", "unit": "s", "value": 9.751381221052725 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[42].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.58129896993453 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[42].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[42].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[42].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.39438293489671017 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[43].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[43].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40510 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[43].ttft_s", "statistic": "single", "unit": "s", "value": 9.746598439989612 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[43].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.30269932980816 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[43].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[43].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[43].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[44].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[44].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60633 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[44].ttft_s", "statistic": "single", "unit": "s", "value": 7.0350889369728975 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[44].decode_tps", "statistic": "single", "unit": "tok/s", "value": 63.24639323212332 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[44].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[44].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[44].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6568040505995085 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[45].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[45].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40509 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[45].ttft_s", "statistic": "single", "unit": "s", "value": 9.609525175008457 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[45].decode_tps", "statistic": "single", "unit": "tok/s", "value": 66.89519669457324 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[45].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[45].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[45].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[46].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[46].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60639 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[46].ttft_s", "statistic": "single", "unit": "s", "value": 7.051564114983194 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[46].decode_tps", "statistic": "single", "unit": "tok/s", "value": 63.54593549511489 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[46].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[46].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[46].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6567390623196293 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[47].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[47].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40507 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[47].ttft_s", "statistic": "single", "unit": "s", "value": 9.608184001001064 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[47].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.48115204779128 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[47].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[47].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[47].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[48].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[48].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60633 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[48].ttft_s", "statistic": "single", "unit": "s", "value": 7.0516899060457945 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[48].decode_tps", "statistic": "single", "unit": "tok/s", "value": 63.62323713692684 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[48].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[48].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[48].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6568040505995085 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[49].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[49].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40513 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[49].ttft_s", "statistic": "single", "unit": "s", "value": 9.641557188995648 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[49].decode_tps", "statistic": "single", "unit": "tok/s", "value": 67.32755393513982 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[49].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[49].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[49].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 9.60064584450447 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 11.872520234028343 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 35 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 7.025318094500108 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 35 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 50 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "49/50" } ], "reasons": [ "49/50 requests passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "de282722e0c5d46bc4f15406dad310be", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "9124cbc7dda94fb8af3860bd", "measured_at": "2026-10-07T16:41:56.510147+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 4.561672913026996 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20196 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 3.9425130789750256 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.1876208907623 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20633 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 4.561672913026996 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.60323666531433 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 4.561672913026996 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 3.9425130789750256 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 3.9425130789750256 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 2 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "2/2" } ], "reasons": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "47a03fd9bbfe612c42f992e302c0e2fe", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "933bd14af8e24e5690f7c9d6", "measured_at": "2026-10-07T21:07:12.591006+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 190.3061074099969 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 50 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20208 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 21.329274443967734 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.800021207463116 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40334 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 25.960546398011502 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.330618235830883 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4665046858729608 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60462 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 48.90890868700808 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.200962357038657 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4794156048060555 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40512 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 74.31105849600863 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.914326809025473 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[3].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3234697077220141 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60640 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[4].ttft_s", "statistic": "single", "unit": "s", "value": 46.11197830497986 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.9343539880554226 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[4].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[4].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.574535299418691 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40504 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[5].ttft_s", "statistic": "single", "unit": "s", "value": 140.01338385697454 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.8171736258173374 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[5].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[5].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.2154899923589008 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60633 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[6].ttft_s", "statistic": "single", "unit": "s", "value": 127.06433388101868 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.3569981933983906 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[6].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[6].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.47855746467487686 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40505 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[7].ttft_s", "statistic": "single", "unit": "s", "value": 174.34134658495896 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.507669802204963 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[7].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[7].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.30613599623057036 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60635 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[8].ttft_s", "statistic": "single", "unit": "s", "value": 136.46487047098344 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.6898082485853099 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[8].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[8].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6464539855042588 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40515 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[9].ttft_s", "statistic": "single", "unit": "s", "value": 174.24829895800212 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.8690146022410477 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[9].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[9].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3058716431497497 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20214 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[10].ttft_s", "statistic": "single", "unit": "s", "value": 48.4673903780058 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.607279068955062 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[10].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[10].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.4858941940601914 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40338 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[11].ttft_s", "statistic": "single", "unit": "s", "value": 46.27357064699754 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.491610196623929 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[11].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[11].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.18618826626030338 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60465 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[12].ttft_s", "statistic": "single", "unit": "s", "value": 38.505369007005356 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.530389796731549 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[12].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[12].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.602165456752806 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40510 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[13].ttft_s", "statistic": "single", "unit": "s", "value": 136.94608946400695 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.5481155264115274 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[13].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[13].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.2154899923589008 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60639 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[14].ttft_s", "statistic": "single", "unit": "s", "value": 129.83873340499122 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.556765745663189 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[14].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[14].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.47855746467487686 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40509 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[15].ttft_s", "statistic": "single", "unit": "s", "value": 174.0081045710249 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.4378033283257645 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[15].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[15].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.30613599623057036 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60642 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[16].ttft_s", "statistic": "single", "unit": "s", "value": 136.43519102502614 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.6381086026939895 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140010 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[16].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[16].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6464539855042588 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40512 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[17].ttft_s", "statistic": "single", "unit": "s", "value": 174.09857321798336 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.8292343589906948 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[17].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[17].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3058716431497497 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60646 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[18].ttft_s", "statistic": "single", "unit": "s", "value": 112.44803817098727 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.545326397213238 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[18].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[18].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6463740395079642 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40514 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[19].ttft_s", "statistic": "single", "unit": "s", "value": 130.2378034510184 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.33629740290711 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[19].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[19].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.528693775709758 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[20].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[20].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20209 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[20].ttft_s", "statistic": "single", "unit": "s", "value": 28.169707986991853 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[20].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.4426631464198238 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[20].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[20].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[20].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5547928726099325 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[21].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[21].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40339 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[21].ttft_s", "statistic": "single", "unit": "s", "value": 35.04971672699321 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[21].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.977797752145614 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[21].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[21].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[21].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.15505179105584535 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[22].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[22].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60471 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[22].ttft_s", "statistic": "single", "unit": "s", "value": 103.26769567304291 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[22].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.496799574675473 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[22].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[22].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[22].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3885150202681943 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[23].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[23].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40507 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[23].ttft_s", "statistic": "single", "unit": "s", "value": 152.32314833800774 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[23].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.199154387630421 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[23].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[23].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[23].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.43749015584146744 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[24].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[24].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60642 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[24].ttft_s", "statistic": "single", "unit": "s", "value": 123.64740463497583 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[24].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.5933244655467504 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[24].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[24].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[24].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6464166749117773 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[25].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[25].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40513 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[25].ttft_s", "statistic": "single", "unit": "s", "value": 189.65001344500342 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[25].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.8664791792833126 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[25].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[25].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[25].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3522653492661513 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[26].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[26].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60645 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[26].ttft_s", "statistic": "single", "unit": "s", "value": 131.7563378829509 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[26].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.5132545744286157 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[26].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[26].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[26].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6463846978316432 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[27].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[27].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40507 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[27].ttft_s", "statistic": "single", "unit": "s", "value": 159.56463948398596 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[27].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.457350469394345 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[27].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[27].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[27].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.21518243847813318 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[28].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[28].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60637 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[28].ttft_s", "statistic": "single", "unit": "s", "value": 109.00755071401363 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[28].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.8370576749938956 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[28].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[28].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[28].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.528693775709758 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[29].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[29].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40513 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[29].ttft_s", "statistic": "single", "unit": "s", "value": 134.29633118602214 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[29].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.703693814203106 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[29].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[29].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[29].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[30].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[30].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20212 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[30].ttft_s", "statistic": "single", "unit": "s", "value": 89.27048925100826 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[30].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.625763886358415 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[30].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[30].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[30].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[31].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[31].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40341 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[31].ttft_s", "statistic": "single", "unit": "s", "value": 129.31412845099112 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[31].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.785944072794717 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[31].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[31].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[31].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.47855746467487686 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[32].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[32].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60478 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[32].ttft_s", "statistic": "single", "unit": "s", "value": 124.47435620002216 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[32].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.3592179293129267 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[32].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[32].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[32].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.44781916635568114 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[33].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[33].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40510 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[33].ttft_s", "statistic": "single", "unit": "s", "value": 189.37543499999447 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[33].decode_tps", "statistic": "single", "unit": "tok/s", "value": 7.205087444041473 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[33].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[33].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[33].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.2767130443376182 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[34].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[34].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60637 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[34].ttft_s", "statistic": "single", "unit": "s", "value": 127.74166052800138 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[34].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.405626634952503 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[34].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[34].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[34].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5286866452674454 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[35].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[35].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40505 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[35].ttft_s", "statistic": "single", "unit": "s", "value": 157.64768635400105 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[35].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.106955563183393 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[35].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[35].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[35].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.21518243847813318 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[36].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[36].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60637 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[36].ttft_s", "statistic": "single", "unit": "s", "value": 109.03757754800608 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[36].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.8623292530691695 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[36].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[36].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[36].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.528693775709758 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[37].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[37].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40518 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[37].ttft_s", "statistic": "single", "unit": "s", "value": 133.9580188699765 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[37].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.71865423723949 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[37].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[37].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[37].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[38].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[38].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60647 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[38].ttft_s", "statistic": "single", "unit": "s", "value": 55.89255939796567 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[38].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.874354797895851 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[38].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[38].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[38].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6463953565068268 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[39].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[39].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40518 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[39].ttft_s", "statistic": "single", "unit": "s", "value": 89.41922640701523 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[39].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.618730323691707 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[39].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[39].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[39].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[40].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[40].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20214 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[40].ttft_s", "statistic": "single", "unit": "s", "value": 43.70142307103379 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[40].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.92405110419344 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[40].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[40].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[40].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[41].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[41].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40342 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[41].ttft_s", "statistic": "single", "unit": "s", "value": 128.91166285204235 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[41].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.657853207894531 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[41].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[41].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[41].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.47855746467487686 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[42].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[42].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60472 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[42].ttft_s", "statistic": "single", "unit": "s", "value": 124.88513929198962 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[42].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.4038742794414685 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[42].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[42].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[42].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.44781916635568114 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[43].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[43].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40514 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[43].ttft_s", "statistic": "single", "unit": "s", "value": 190.3061074099969 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[43].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.770244051459168 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[43].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[43].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[43].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3058692564984212 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[44].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[44].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60641 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[44].ttft_s", "statistic": "single", "unit": "s", "value": 134.05116806202568 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[44].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.8572560243103462 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[44].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[44].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[44].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.38754708400478505 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[45].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[45].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40513 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[45].ttft_s", "statistic": "single", "unit": "s", "value": 158.35320276900893 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[45].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.776806052317717 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[45].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[45].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[45].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.38750877331725303 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[46].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[46].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60647 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[46].ttft_s", "statistic": "single", "unit": "s", "value": 110.68467743898509 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[46].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.0350651155027744 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[46].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[46].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[46].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.528693775709758 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[47].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[47].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40511 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[47].ttft_s", "statistic": "single", "unit": "s", "value": 133.19303036900237 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[47].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.152866995766637 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[47].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[47].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[47].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[48].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[48].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60641 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[48].ttft_s", "statistic": "single", "unit": "s", "value": 56.41038528998615 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[48].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.84782894209172 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[48].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[48].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[48].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6463953565068268 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[49].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[49].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40517 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[49].ttft_s", "statistic": "single", "unit": "s", "value": 88.89149994001491 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[49].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.473663996997434 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[49].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[49].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[49].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 154.9854173460044 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 136.46487047098344 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 35 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 109.02256413100986 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 35 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 50 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "50/50" } ], "reasons": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "7019c1aa88ec899aabef8d10e6251014", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Measured retrieval only, with no customer qualification threshold. Exact dense float32 array size excludes payloads and vector database overhead. One timing run; no reranking, ingestion, answer generation or GPU measurement.", "Publisher revision was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "SentenceTransformers CPU", "recipe": "plain", "version": "6.1.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "96f87869701b4783ba4bf5d6", "measured_at": "2026-10-07T14:37:01.394807+00:00", "metrics": [ { "basis": null, "n": null, "name": "passages_per_second", "scope": "machine_total", "source_field": "passages_per_second", "statistic": "single", "unit": "passages/s", "value": 1.6377562878110878 }, { "basis": null, "n": null, "name": "query_seconds", "scope": "per_person", "source_field": "query_seconds", "statistic": "single", "unit": "s", "value": 6.8647271250374615 }, { "basis": null, "n": null, "name": "recall_at_5", "scope": "machine_total", "source_field": "recall_at_5", "statistic": "single", "unit": "ratio", "value": 0.9 }, { "basis": null, "n": null, "name": "mrr", "scope": "machine_total", "source_field": "mrr", "statistic": "single", "unit": "ratio", "value": 0.7899252411134435 }, { "basis": null, "n": null, "name": "questions", "scope": "machine_total", "source_field": "questions", "statistic": "single", "unit": "count", "value": 100 }, { "basis": null, "n": null, "name": "passage_seconds", "scope": "machine_total", "source_field": "passage_seconds", "statistic": "single", "unit": "s", "value": 244.23658329201862 }, { "basis": null, "n": null, "name": "vector_bytes", "scope": "machine_total", "source_field": "vector_bytes", "statistic": "single", "unit": "bytes", "value": 1638400 }, { "basis": null, "n": null, "name": "passages", "scope": "machine_total", "source_field": "passages", "statistic": "single", "unit": "count", "value": 400 }, { "basis": null, "n": null, "name": "index_file_bytes", "scope": "machine_total", "source_field": "index_file_bytes", "statistic": "single", "unit": "bytes", "value": 1638528 }, { "basis": null, "n": null, "name": "recall_at_10", "scope": "machine_total", "source_field": "recall_at_10", "statistic": "single", "unit": "ratio", "value": 0.94 }, { "basis": null, "n": null, "name": "truncated_passages", "scope": "machine_total", "source_field": "truncated_passages", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-0.6B-Base", "publisher_revision": null, "variant": "FP32 variant", "weights_dtype": "FP32", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-embedding-eval-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "embeddings", "version": "1" }, "settings": { "context_tokens": 32768, "corpus_sha256": "d7fb630b25a7d5128c44491d1a8f21bde479711f79520f1ce95ea8bd8a0ae1f6", "dimensions": 1024, "dtype": "float32" }, "setup_complete": false, "setup_id": "bc676c0fedd62924fbfb726dc83eacba", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple Silicon laptop CPU, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "a3fe8bb136e546e19df1a3e6", "measured_at": "2026-10-07T01:54:48.447482+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 644.1904842269723 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 633.7372273710207 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 636.8561760049779 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 633.7372273710207 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 643.4930662060506 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.020239211416886 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 394.50262538172046 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 89 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[0].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 633.7315072939964 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 608.4416262260056 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 643.7070903419517 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.8072790896642386 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 410.9038389610993 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 100 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[1].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 608.4403012630064 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 626.0772230169969 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 643.9363833609968 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.487380039841819 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 399.3293332014614 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 99 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[2].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 626.0738319699885 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 636.3248575759935 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 644.1460858180071 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.53000129488224 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 392.8983710496368 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 99 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[3].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 636.3144952320145 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 636.8561760049779 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 644.1812003189698 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.378795181990778 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 392.5705825266988 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 99 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[4].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 636.844163976959 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 13.458033408947514 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.7544352360050759 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 100.8923454284668 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 636.8561760049779 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 9.020239211416886, "effective_prefill_tps": 394.50262538172046, "first_real_token_from_dispatch_s": 633.7315072939964, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 89, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 643.4930662060506, "ttft_s": 633.7372273710207, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 2.8072790896642386, "effective_prefill_tps": 410.9038389610993, "first_real_token_from_dispatch_s": 608.4403012630064, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 100, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 643.7070903419517, "ttft_s": 608.4416262260056, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.487380039841819, "effective_prefill_tps": 399.3293332014614, "first_real_token_from_dispatch_s": 626.0738319699885, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 99, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 643.9363833609968, "ttft_s": 626.0772230169969, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 12.53000129488224, "effective_prefill_tps": 392.8983710496368, "first_real_token_from_dispatch_s": 636.3144952320145, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 99, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 644.1460858180071, "ttft_s": 636.3248575759935, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 13.378795181990778, "effective_prefill_tps": 392.5705825266988, "first_real_token_from_dispatch_s": 636.844163976959, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 99, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 644.1812003189698, "ttft_s": 636.8561760049779, "user_number": 2, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "1455fb8523cb693bc00de6e3284af4dd", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Operator-supplied runtime manifest; the endpoint runner cannot attest hardware or model pins. Bounded task correctness is not customer acceptance.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "a8379842455c44d3a0045dbf", "measured_at": "2026-10-07T04:19:42.894630+00:00", "metrics": [ { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 26 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[0].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[0].passed", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[0].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[1].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[1].passed", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[1].total", "statistic": "single", "unit": "count", "value": 16 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "structured-tasks", "observed": "26/32" } ], "reasons": [ "26/32 tasks passed." ], "status": "partial" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 26, "skipped": null, "suite": "structured-tasks", "total": 32, "version": null } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": { "context_tokens": 262144, "kv_dtype": "fp8", "speculative_decoding": "plain" }, "setup_complete": false, "setup_id": "78643e8f3a489b3b19871370e5162fcd", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "a95b7a20023446a4950ba9fc", "measured_at": "2026-10-07T16:57:11.795777+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 4.744170687976293 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20195 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 3.956520487961825 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.92648374257273 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20794 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 4.172652253997512 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.63057496321937 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21323 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 4.337529264972545 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.1574084308641 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21728 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 4.4090711689786986 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.34338854648301 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[3].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21724 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[4].ttft_s", "statistic": "single", "unit": "s", "value": 4.401588754961267 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.44802875165968 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[4].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[4].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21727 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[5].ttft_s", "statistic": "single", "unit": "s", "value": 4.410432644013781 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.09937869880784 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[5].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[5].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21725 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[6].ttft_s", "statistic": "single", "unit": "s", "value": 4.404367696959525 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.39726802014195 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[6].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[6].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21724 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[7].ttft_s", "statistic": "single", "unit": "s", "value": 4.421268408012111 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.08454459870406 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[7].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[7].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21725 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[8].ttft_s", "statistic": "single", "unit": "s", "value": 4.409595528035425 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.4323385281416 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[8].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[8].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21729 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[9].ttft_s", "statistic": "single", "unit": "s", "value": 4.744170687976293 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.37303676080445 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[9].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[9].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20201 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[10].ttft_s", "statistic": "single", "unit": "s", "value": 3.9617760230321437 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.46756188121972 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[10].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[10].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20784 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[11].ttft_s", "statistic": "single", "unit": "s", "value": 4.164171846001409 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.71260643544663 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[11].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[11].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21311 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[12].ttft_s", "statistic": "single", "unit": "s", "value": 4.340403509035241 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.41988530149786 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[12].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[12].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21708 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[13].ttft_s", "statistic": "single", "unit": "s", "value": 4.411751301027834 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.63941646912875 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[13].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[13].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21709 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[14].ttft_s", "statistic": "single", "unit": "s", "value": 4.411505792988464 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 70.95477128557425 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[14].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[14].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21705 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[15].ttft_s", "statistic": "single", "unit": "s", "value": 4.403040006989613 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.15907973193424 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[15].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[15].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21709 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[16].ttft_s", "statistic": "single", "unit": "s", "value": 4.39987631700933 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.14043649360671 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140010 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[16].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[16].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21707 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[17].ttft_s", "statistic": "single", "unit": "s", "value": 4.40034231095342 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.16683607203849 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[17].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[17].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21707 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[18].ttft_s", "statistic": "single", "unit": "s", "value": 4.430036486999597 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.04510288736297 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[18].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[18].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 21709 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[19].ttft_s", "statistic": "single", "unit": "s", "value": 4.417829783982597 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 70.85562286778652 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[19].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[19].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 4.406719432969112 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 3.9617760230321437 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 3.9591482554969843 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 20 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "12/20" } ], "reasons": [ "12/20 requests passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 2, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "47a03fd9bbfe612c42f992e302c0e2fe", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. No summaries. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "ac1380ca0c6a4717af067bec", "measured_at": "2026-10-07T16:29:41.609051+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 4.051701668009628 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20195 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 5.51429703400936 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.18935760740781 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20255 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 4.010435424046591 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.85426650567737 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20256 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 4.011913746013306 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.90258261947046 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20258 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 4.051701668009628 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.03965771803414 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[3].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20254 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[4].ttft_s", "statistic": "single", "unit": "s", "value": 4.01388861896703 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.06988707964224 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[4].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[4].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20257 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[5].ttft_s", "statistic": "single", "unit": "s", "value": 4.026952326006722 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.64814115349647 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[5].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[5].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20255 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[6].ttft_s", "statistic": "single", "unit": "s", "value": 4.028189387056045 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.7489600228354 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[6].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[6].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20254 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[7].ttft_s", "statistic": "single", "unit": "s", "value": 4.035180857987143 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.7480638711258 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[7].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[7].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20255 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[8].ttft_s", "statistic": "single", "unit": "s", "value": 4.018719704006799 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.27279824942455 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[8].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[8].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20259 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[9].ttft_s", "statistic": "single", "unit": "s", "value": 4.020340268034488 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.0115461878576 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[9].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[9].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20201 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[10].ttft_s", "statistic": "single", "unit": "s", "value": 3.9575468280236237 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.18914243591527 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[10].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[10].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20257 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[11].ttft_s", "statistic": "single", "unit": "s", "value": 4.01160303899087 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.80232916002316 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[11].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[11].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20257 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[12].ttft_s", "statistic": "single", "unit": "s", "value": 4.018064144998789 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.14069700933722 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[12].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[12].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20260 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[13].ttft_s", "statistic": "single", "unit": "s", "value": 4.020838255994022 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.30027471918027 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[13].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[13].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20261 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[14].ttft_s", "statistic": "single", "unit": "s", "value": 4.020210571005009 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.07412454164407 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[14].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[14].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20257 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[15].ttft_s", "statistic": "single", "unit": "s", "value": 4.022851930989418 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.91559659430202 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[15].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[15].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20261 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[16].ttft_s", "statistic": "single", "unit": "s", "value": 4.020499258011114 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.73848015792636 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140010 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[16].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[16].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20259 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[17].ttft_s", "statistic": "single", "unit": "s", "value": 4.023691577021964 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.80952433247298 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[17].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[17].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20259 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[18].ttft_s", "statistic": "single", "unit": "s", "value": 4.032263829023577 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.92103241900539 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[18].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[18].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20261 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[19].ttft_s", "statistic": "single", "unit": "s", "value": 4.0381233529769816 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.91700831048671 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[19].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[19].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[20].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[20].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20196 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[20].ttft_s", "statistic": "single", "unit": "s", "value": 3.9496893669711426 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[20].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.84703388538237 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[20].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[20].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[20].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[21].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[21].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20257 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[21].ttft_s", "statistic": "single", "unit": "s", "value": 4.007999450026546 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[21].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.27014438670055 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[21].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[21].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[21].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[22].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[22].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20259 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[22].ttft_s", "statistic": "single", "unit": "s", "value": 4.021685317973606 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[22].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.02844945983882 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[22].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[22].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[22].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[23].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[23].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20256 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[23].ttft_s", "statistic": "single", "unit": "s", "value": 4.0203455240116455 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[23].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.83655179960931 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[23].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[23].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[23].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[24].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[24].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20259 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[24].ttft_s", "statistic": "single", "unit": "s", "value": 4.019376400974579 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[24].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.44322510344456 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[24].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[24].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[24].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[25].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[25].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20256 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[25].ttft_s", "statistic": "single", "unit": "s", "value": 4.013743037998211 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[25].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.90022278246158 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[25].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[25].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[25].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[26].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[26].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20256 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[26].ttft_s", "statistic": "single", "unit": "s", "value": 4.015660401026253 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[26].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.96538149551826 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[26].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[26].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[26].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[27].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[27].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20256 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[27].ttft_s", "statistic": "single", "unit": "s", "value": 4.015170220984146 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[27].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.70720114519966 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[27].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[27].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[27].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[28].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[28].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20256 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[28].ttft_s", "statistic": "single", "unit": "s", "value": 4.012479936995078 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[28].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.87884088980499 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[28].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[28].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[28].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[29].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[29].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20261 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[29].ttft_s", "statistic": "single", "unit": "s", "value": 4.0447196930181235 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[29].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.89135676960117 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[29].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[29].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[29].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[30].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[30].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20199 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[30].ttft_s", "statistic": "single", "unit": "s", "value": 3.9469772179727443 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[30].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.7190664192229 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[30].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[30].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[30].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[31].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[31].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20259 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[31].ttft_s", "statistic": "single", "unit": "s", "value": 4.010930905002169 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[31].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.20883328515136 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[31].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[31].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[31].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[32].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[32].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20262 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[32].ttft_s", "statistic": "single", "unit": "s", "value": 4.022729684016667 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[32].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.00766179961506 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[32].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[32].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[32].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[33].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[33].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20259 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[33].ttft_s", "statistic": "single", "unit": "s", "value": 4.010955834994093 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[33].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.90837025297432 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[33].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[33].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[33].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[34].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[34].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20258 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[34].ttft_s", "statistic": "single", "unit": "s", "value": 4.013563547981903 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[34].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.10719468868285 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[34].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[34].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[34].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[35].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[35].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20259 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[35].ttft_s", "statistic": "single", "unit": "s", "value": 4.022994794009719 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[35].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.14818572878679 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[35].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[35].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[35].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[36].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[36].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20260 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[36].ttft_s", "statistic": "single", "unit": "s", "value": 4.016720690007787 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[36].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.2866153377783 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[36].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[36].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[36].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[37].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[37].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20261 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[37].ttft_s", "statistic": "single", "unit": "s", "value": 4.0264908179524355 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[37].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.57903423661371 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[37].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[37].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[37].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[38].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[38].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20259 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[38].ttft_s", "statistic": "single", "unit": "s", "value": 4.026787961018272 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[38].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.4769349447062 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[38].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[38].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[38].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[39].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[39].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20259 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[39].ttft_s", "statistic": "single", "unit": "s", "value": 4.040196919988375 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[39].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.93154239902539 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[39].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[39].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[39].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[40].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[40].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20201 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[40].ttft_s", "statistic": "single", "unit": "s", "value": 3.957647190953139 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[40].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.33642584779359 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[40].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[40].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[40].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[41].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[41].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20259 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[41].ttft_s", "statistic": "single", "unit": "s", "value": 4.0205346629954875 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[41].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.75451466690026 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[41].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[41].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[41].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[42].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[42].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20260 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[42].ttft_s", "statistic": "single", "unit": "s", "value": 4.024785568006337 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[42].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.66623410345335 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[42].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[42].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[42].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[43].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[43].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20258 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[43].ttft_s", "statistic": "single", "unit": "s", "value": 4.036519845016301 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[43].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.91822700413185 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[43].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[43].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[43].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[44].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[44].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20258 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[44].ttft_s", "statistic": "single", "unit": "s", "value": 4.033242037054151 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[44].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.7765869196085 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[44].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[44].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[44].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[45].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[45].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20259 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[45].ttft_s", "statistic": "single", "unit": "s", "value": 4.036150368978269 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[45].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.80101458615283 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[45].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[45].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[45].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[46].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[46].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20259 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[46].ttft_s", "statistic": "single", "unit": "s", "value": 4.032587441965006 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[46].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.74616165833895 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[46].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[46].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[46].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[47].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[47].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20258 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[47].ttft_s", "statistic": "single", "unit": "s", "value": 4.0359164659748785 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[47].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.76507202583112 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[47].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[47].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[47].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[48].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[48].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20259 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[48].ttft_s", "statistic": "single", "unit": "s", "value": 4.039411919016857 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[48].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.71539436549808 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[48].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[48].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[48].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[49].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[49].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20264 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[49].ttft_s", "statistic": "single", "unit": "s", "value": 4.04769021901302 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[49].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.59676937055349 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[49].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[49].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[49].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 4.021685317973606 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 45 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 5.51429703400936 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 3.9575468280236237 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 45 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 50 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "5/50" } ], "reasons": [ "5/50 requests passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "47a03fd9bbfe612c42f992e302c0e2fe", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "b043ed1be0a3435aa8b429e9", "measured_at": "2026-10-07T19:05:00.468478+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 4.092750951007474 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 42 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20208 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 3.9518823719699867 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.88899221614913 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20377 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 4.044749770022463 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.75192220846411 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 4.052152278949507 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.06284040053211 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 4.064480888016988 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.66132274861248 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[3].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20377 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[4].ttft_s", "statistic": "single", "unit": "s", "value": 4.060537749959622 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.6908034851705 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[4].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[4].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20377 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[5].ttft_s", "statistic": "single", "unit": "s", "value": 4.062791217991617 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.45478611745676 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[5].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[5].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[6].ttft_s", "statistic": "single", "unit": "s", "value": 4.0748625620035455 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.36262575236603 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[6].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[6].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20378 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[7].ttft_s", "statistic": "single", "unit": "s", "value": 4.062875127012376 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.00463960044767 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[7].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[7].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[8].ttft_s", "statistic": "single", "unit": "s", "value": 4.069741029001307 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.29031856342269 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[8].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[8].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20384 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[9].ttft_s", "statistic": "single", "unit": "s", "value": 4.084067455027252 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.11800564181584 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[9].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[9].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20214 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[10].ttft_s", "statistic": "single", "unit": "s", "value": 3.970529598998837 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.59190954757892 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[10].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[10].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[11].ttft_s", "statistic": "single", "unit": "s", "value": 4.076273841958027 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.82220695617184 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[11].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[11].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[12].ttft_s", "statistic": "single", "unit": "s", "value": 4.0787452919757925 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.80642482069852 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[12].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[12].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[13].ttft_s", "statistic": "single", "unit": "s", "value": 4.065320292022079 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.89650100395845 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[13].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[13].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[14].ttft_s", "statistic": "single", "unit": "s", "value": 4.0545590440160595 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.65278197204479 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[14].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[14].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[15].ttft_s", "statistic": "single", "unit": "s", "value": 4.063409916008823 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.98973093280988 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[15].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[15].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20384 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[16].ttft_s", "statistic": "single", "unit": "s", "value": 4.06304108502809 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.48201637031023 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140010 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[16].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[16].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[17].ttft_s", "statistic": "single", "unit": "s", "value": 4.072712241963018 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.63821856214857 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[17].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[17].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[18].ttft_s", "statistic": "single", "unit": "s", "value": 4.078892498044297 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.67098998010962 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[18].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[18].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[19].ttft_s", "statistic": "single", "unit": "s", "value": 4.071676758991089 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.58385573767384 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[19].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[19].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[20].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[20].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20209 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[20].ttft_s", "statistic": "single", "unit": "s", "value": 3.967386535019614 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[20].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.2303146909077 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[20].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[20].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[20].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[21].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[21].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[21].ttft_s", "statistic": "single", "unit": "s", "value": 4.048619043023791 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[21].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.18794778830382 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[21].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[21].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[21].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[22].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[22].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[22].ttft_s", "statistic": "single", "unit": "s", "value": 4.052427285991143 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[22].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.06183075646165 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[22].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[22].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[22].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[23].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[23].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20378 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[23].ttft_s", "statistic": "single", "unit": "s", "value": 4.044971527997404 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[23].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.08472077377472 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[23].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[23].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[23].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[24].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[24].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[24].ttft_s", "statistic": "single", "unit": "s", "value": 4.0471543680178 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[24].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.77278305431133 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[24].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[24].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[24].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[25].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[25].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[25].ttft_s", "statistic": "single", "unit": "s", "value": 4.054669192992151 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[25].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.86991972338258 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[25].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[25].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[25].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[26].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[26].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[26].ttft_s", "statistic": "single", "unit": "s", "value": 4.048213255009614 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[26].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.80461971092599 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[26].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[26].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[26].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[27].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[27].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[27].ttft_s", "statistic": "single", "unit": "s", "value": 4.06273244402837 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[27].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.67068639844285 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[27].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[27].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[27].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[28].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[28].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[28].ttft_s", "statistic": "single", "unit": "s", "value": 4.079237227968406 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[28].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.92857383710039 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[28].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[28].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[28].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[29].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[29].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[29].ttft_s", "statistic": "single", "unit": "s", "value": 4.082743456005119 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[29].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.61795224538622 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[29].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[29].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[29].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[30].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[30].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20212 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[30].ttft_s", "statistic": "single", "unit": "s", "value": 3.968744986981619 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[30].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.26185035315635 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[30].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[30].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[30].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[31].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[31].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[31].ttft_s", "statistic": "single", "unit": "s", "value": 4.048114881967194 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[31].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.0338979998823 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[31].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[31].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[31].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[32].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[32].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20385 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[32].ttft_s", "statistic": "single", "unit": "s", "value": 4.045793760975357 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[32].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.60253730387652 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[32].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[32].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[32].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[33].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[33].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[33].ttft_s", "statistic": "single", "unit": "s", "value": 4.054431747994386 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[33].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.01643941196475 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[33].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[33].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[33].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[34].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[34].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20376 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[34].ttft_s", "statistic": "single", "unit": "s", "value": 4.060337501985487 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[34].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.38486391880528 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[34].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[34].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[34].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[35].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[35].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[35].ttft_s", "statistic": "single", "unit": "s", "value": 4.052846778999083 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[35].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.23048516524847 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[35].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[35].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[35].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[36].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[36].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[36].ttft_s", "statistic": "single", "unit": "s", "value": 4.0723483199835755 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[36].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.98136642138937 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[36].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[36].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[36].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[37].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[37].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20385 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[37].ttft_s", "statistic": "single", "unit": "s", "value": 4.049312277988065 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[37].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.0497620416116 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[37].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[37].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[37].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[38].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[38].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20383 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[38].ttft_s", "statistic": "single", "unit": "s", "value": 4.068096189992502 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[38].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.00481463206478 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[38].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[38].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[38].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[39].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[39].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20384 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[39].ttft_s", "statistic": "single", "unit": "s", "value": 4.054706171038561 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[39].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.32012000702774 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[39].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[39].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[39].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[40].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[40].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20214 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[40].ttft_s", "statistic": "single", "unit": "s", "value": 3.966890710988082 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[40].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.14467708224605 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[40].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[40].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[40].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[41].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[41].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[41].ttft_s", "statistic": "single", "unit": "s", "value": 4.066985448007472 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[41].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.78913392952829 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[41].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[41].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[41].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[42].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[42].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[42].ttft_s", "statistic": "single", "unit": "s", "value": 4.050281835952774 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[42].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.49755047706364 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[42].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[42].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[42].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[43].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[43].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[43].ttft_s", "statistic": "single", "unit": "s", "value": 4.0613198800128885 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[43].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.55899888745996 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[43].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[43].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[43].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[44].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[44].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20379 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[44].ttft_s", "statistic": "single", "unit": "s", "value": 4.058420856017619 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[44].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.62791577326144 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[44].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[44].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[44].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[45].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[45].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[45].ttft_s", "statistic": "single", "unit": "s", "value": 4.071163663989864 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[45].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.93700640736388 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[45].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[45].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[45].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[46].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[46].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20381 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[46].ttft_s", "statistic": "single", "unit": "s", "value": 4.063698283047415 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[46].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.73045891945799 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[46].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[46].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[46].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[47].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[47].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20382 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[47].ttft_s", "statistic": "single", "unit": "s", "value": 4.0701730260043405 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[47].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.53307418681935 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[47].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[47].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[47].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[48].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[48].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[48].ttft_s", "statistic": "single", "unit": "s", "value": 4.080761422985233 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[48].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.39699198742929 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[48].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[48].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[48].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[49].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[49].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20386 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[49].ttft_s", "statistic": "single", "unit": "s", "value": 4.092750951007474 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[49].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.40463445088477 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[49].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[49].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[49].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 4.062875127012376 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 3.970529598998837 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 37 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 3.967386535019614 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 45 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 50 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "42/50" } ], "reasons": [ "42/50 requests passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "47a03fd9bbfe612c42f992e302c0e2fe", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. No summaries. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "b2acdc4b13a24c219a0f2641", "measured_at": "2026-10-07T15:55:39.656970+00:00", "metrics": [ { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60001 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120006 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140006 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200009 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60002 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100007 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120007 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140010 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160011 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180012 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200014 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 50 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/50" } ], "reasons": [ "0/50 requests passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "75b3e85e01b0e1f32d60584d2ee930b3", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "b5aac3fc0a9047179b220b3c", "measured_at": "2026-10-07T17:22:54.826282+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 4.194312653038651 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20751 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 4.108138862007763 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.11937128972566 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20808 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 4.177831583016086 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.90830654578947 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20809 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 4.162368046003394 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.82029660112212 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20811 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 4.176804709946737 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.88372443392322 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[3].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20807 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[4].ttft_s", "statistic": "single", "unit": "s", "value": 4.173004061973188 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.7062503784686 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[4].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[4].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20810 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[5].ttft_s", "statistic": "single", "unit": "s", "value": 4.179843931982759 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.48932624516645 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[5].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[5].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20808 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[6].ttft_s", "statistic": "single", "unit": "s", "value": 4.162876238988247 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.54746542811947 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[6].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[6].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20807 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[7].ttft_s", "statistic": "single", "unit": "s", "value": 4.165472546010278 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.5358376702462 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[7].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[7].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20808 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[8].ttft_s", "statistic": "single", "unit": "s", "value": 4.185213335964363 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.67262028384887 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[8].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[8].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20812 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[9].ttft_s", "statistic": "single", "unit": "s", "value": 4.173682583030313 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.58833336722687 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[9].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[9].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20745 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[10].ttft_s", "statistic": "single", "unit": "s", "value": 4.097530459985137 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.95276992821607 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[10].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[10].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20798 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[11].ttft_s", "statistic": "single", "unit": "s", "value": 4.1694475439726375 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.91391530597619 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[11].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[11].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20798 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[12].ttft_s", "statistic": "single", "unit": "s", "value": 4.169935983023606 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.88902241977326 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[12].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[12].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20801 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[13].ttft_s", "statistic": "single", "unit": "s", "value": 4.167277147003915 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.24259536907466 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[13].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[13].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20802 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[14].ttft_s", "statistic": "single", "unit": "s", "value": 4.17980104801245 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.88067653060499 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[14].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[14].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20798 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[15].ttft_s", "statistic": "single", "unit": "s", "value": 4.188346009992529 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.89953572837202 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[15].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[15].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20802 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[16].ttft_s", "statistic": "single", "unit": "s", "value": 4.179852842993569 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.73474029100784 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140010 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[16].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[16].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20800 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[17].ttft_s", "statistic": "single", "unit": "s", "value": 4.181811859947629 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.43947353880048 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[17].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[17].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20800 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[18].ttft_s", "statistic": "single", "unit": "s", "value": 4.184096890967339 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.65605267201184 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[18].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[18].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20802 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[19].ttft_s", "statistic": "single", "unit": "s", "value": 4.194312653038651 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.36329822572546 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[19].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[19].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 4.1773181464814115 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 4.108138862007763 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 4.10283466099645 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 20 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "12/20" } ], "reasons": [ "12/20 requests passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 2, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "47a03fd9bbfe612c42f992e302c0e2fe", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp CUDA", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "b911e8d530594210b2fc3d02", "measured_at": "2026-10-07T01:06:34.322344+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 629.5073147419607 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 375.66302481101593 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 624.6775280639995 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 127.27466071100207 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 491.53372020000825 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.24433160324097894 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 1964.38158705999 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 90 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[0].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 127.26417148596374 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 249.8476865470293 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 624.6742157849949 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.2401110726686633 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 1000.6736642443916 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 91 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[1].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 249.83982097601984 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 375.66302481101593 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 626.8256207300001 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.3503706420855181 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 665.5326276142696 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 89 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[2].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 375.6497323960066 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 501.0691677770228 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 628.4859969120007 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.6906466013745953 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 498.9650453034017 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 89 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[3].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 501.0680842609727 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 624.6775280639995 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 629.500791908009 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 18.452235431934398 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 400.23210179314367 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250016 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 90 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[4].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 624.672843057022 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 0.884064732178223 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.7132562076487516 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 36.17306137084961 } ], "model": { "artifact_sha256": [], "publisher_repo": "nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16", "publisher_revision": "a9904d24bcc1d289a1950fa9d2b978c47cf903b9", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 624.6775280639995 } ], "reasons": [ "Observed decode overlap was 3 of 5 requested users." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "04cf656671acc024178598a26462c2de38b6f97137ea5abf44a09b83b8047c5e", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.24433160324097894, "effective_prefill_tps": 1964.38158705999, "first_real_token_from_dispatch_s": 127.26417148596374, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 90, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": true, "total_s": 491.53372020000825, "ttft_s": 127.27466071100207, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.2401110726686633, "effective_prefill_tps": 1000.6736642443916, "first_real_token_from_dispatch_s": 249.83982097601984, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 91, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": true, "total_s": 624.6742157849949, "ttft_s": 249.8476865470293, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.3503706420855181, "effective_prefill_tps": 665.5326276142696, "first_real_token_from_dispatch_s": 375.6497323960066, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 89, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": true, "total_s": 626.8256207300001, "ttft_s": 375.66302481101593, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 0.6906466013745953, "effective_prefill_tps": 498.9650453034017, "first_real_token_from_dispatch_s": 501.0680842609727, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 89, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": true, "total_s": 628.4859969120007, "ttft_s": 501.0691677770228, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 18.452235431934398, "effective_prefill_tps": 400.23210179314367, "first_real_token_from_dispatch_s": 624.672843057022, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 90, "prompt_tokens": 250016, "stop_and_done": true, "strict_json": true, "total_s": 629.500791908009, "ttft_s": 624.6775280639995, "user_number": 0, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "q8_0", "max_output_tokens": 512, "scenario": "distinct", "thinking": false }, "setup_complete": false, "setup_id": "9e5118825b2267598d5691416d3ab69e", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250016, "median": 250016, "min": 250016 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Lab retrieval and rolling-history evidence only. No summaries. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "c4e683141a094897a2b3092b", "measured_at": "2026-10-07T11:12:04.658512+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 879.4844111460261 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 60 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20113 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 21.54114813398337 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.922366602566832 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40242 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 25.688145256019197 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.404503784871029 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.46757119427463845 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60372 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 48.5354347829707 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.256194217884328 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.48050754105964105 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80504 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 54.79522712499602 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.898233512725842 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[3].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7401371360429295 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100636 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[4].ttft_s", "statistic": "single", "unit": "s", "value": 81.66109736502403 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.831066147361787 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[4].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[4].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6579069407945587 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120764 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[5].ttft_s", "statistic": "single", "unit": "s", "value": 90.72225817199796 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.4919241224625357 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[5].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[5].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6831878430233447 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 140892 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[6].ttft_s", "statistic": "single", "unit": "s", "value": 108.68508133402793 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.638065034169022 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140015 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[6].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[6].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.714141784968022 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 161021 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[7].ttft_s", "statistic": "single", "unit": "s", "value": 153.7950589460088 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.847934177573787 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160018 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[7].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[7].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7722619569655089 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181154 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[8].ttft_s", "statistic": "single", "unit": "s", "value": 174.2152354019927 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.990309555166853 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[8].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[8].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8151799370459877 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201288 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[9].ttft_s", "statistic": "single", "unit": "s", "value": 190.19731120503275 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.777932223310172 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[9].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[9].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8427175871771364 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221414 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[10].ttft_s", "statistic": "single", "unit": "s", "value": 201.21917534497334 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.671131467089102 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[10].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[10].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8650683874881357 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241548 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[11].ttft_s", "statistic": "single", "unit": "s", "value": 218.34677868604194 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.970563389721024 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[11].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[11].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8804022329845835 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181158 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[12].ttft_s", "statistic": "single", "unit": "s", "value": 675.7934919499676 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.908612530750501 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[12].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[12].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8189083824090568 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201294 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[13].ttft_s", "statistic": "single", "unit": "s", "value": 209.02731904800748 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.3801402778578878 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[13].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[13].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.47146674475639666 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221419 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[14].ttft_s", "statistic": "single", "unit": "s", "value": 219.3602335259784 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.8734765494972176 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[14].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[14].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9063798703681333 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241551 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[15].ttft_s", "statistic": "single", "unit": "s", "value": 226.72736896999413 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.37650507741443423 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[15].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[15].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.757284447151076 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181159 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[16].ttft_s", "statistic": "single", "unit": "s", "value": 879.4844111460261 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.3025797107633407 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[16].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[16].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5679568043736538 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201288 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[17].ttft_s", "statistic": "single", "unit": "s", "value": 161.02841810602695 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.2939384819953723 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[17].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[17].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6971398765539425 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221416 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[18].ttft_s", "statistic": "single", "unit": "s", "value": 160.0581362210214 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.4649815194263156 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[18].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[18].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9063757771723853 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241544 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[19].ttft_s", "statistic": "single", "unit": "s", "value": 113.99317735101795 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.23556615332464062 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[19].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[19].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7572788740168346 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[20].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[20].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181142 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[20].ttft_s", "statistic": "single", "unit": "s", "value": 775.65689926903 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[20].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.2412350162140507 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[20].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[20].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[20].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[20].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5679517103059916 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[21].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[21].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20120 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[21].ttft_s", "statistic": "single", "unit": "s", "value": 47.92592330498155 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[21].decode_tps", "statistic": "single", "unit": "tok/s", "value": 11.676440534830242 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[21].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[21].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[21].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[21].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.487004919743577 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[22].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[22].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40247 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[22].ttft_s", "statistic": "single", "unit": "s", "value": 53.98053695401177 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[22].decode_tps", "statistic": "single", "unit": "tok/s", "value": 8.530757313262535 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[22].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[22].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40005 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[22].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[22].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6492699853417363 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[23].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[23].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60378 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[23].ttft_s", "statistic": "single", "unit": "s", "value": 80.98764683300396 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[23].decode_tps", "statistic": "single", "unit": "tok/s", "value": 8.879300057073015 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[23].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[23].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[23].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[23].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6579069407945587 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[24].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[24].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80511 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[24].ttft_s", "statistic": "single", "unit": "s", "value": 91.74500297900522 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[24].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.652123285210395 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[24].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[24].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[24].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[24].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6831878430233447 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[25].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[25].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100640 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[25].ttft_s", "statistic": "single", "unit": "s", "value": 114.38866522099124 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[25].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.857620661717785 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[25].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[25].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[25].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[25].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.714141784968022 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[26].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[26].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120767 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[26].ttft_s", "statistic": "single", "unit": "s", "value": 156.64834343799157 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[26].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.480504584066381 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[26].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[26].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[26].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[26].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7722619569655089 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[27].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[27].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 140899 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[27].ttft_s", "statistic": "single", "unit": "s", "value": 173.25792252796236 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[27].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.966854682480513 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[27].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[27].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140015 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[27].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[27].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8151799370459877 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[28].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[28].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 161030 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[28].ttft_s", "statistic": "single", "unit": "s", "value": 188.31956962001277 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[28].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.9564960080441836 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[28].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[28].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160019 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[28].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[28].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8427175871771364 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[29].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[29].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181163 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[29].ttft_s", "statistic": "single", "unit": "s", "value": 202.2053211479797 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[29].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.330562592184002 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[29].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[29].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[29].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[29].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8650683874881357 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[30].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[30].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201295 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[30].ttft_s", "statistic": "single", "unit": "s", "value": 217.2184149249806 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[30].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.970425881762146 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[30].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[30].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[30].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[30].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8804022329845835 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[31].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[31].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221423 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[31].ttft_s", "statistic": "single", "unit": "s", "value": 199.4954831119976 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[31].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.344248744098066 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[31].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[31].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[31].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[31].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7701590784421284 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[32].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[32].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241560 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[32].ttft_s", "statistic": "single", "unit": "s", "value": 223.942234342976 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[32].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.623770107732449 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[32].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[32].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[32].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[32].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9152508693492301 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[33].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[33].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181172 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[33].ttft_s", "statistic": "single", "unit": "s", "value": 705.0862642640132 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[33].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.687095864929719 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[33].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[33].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[33].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[33].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8140977464693939 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[34].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[34].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201305 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[34].ttft_s", "statistic": "single", "unit": "s", "value": 230.6080427240231 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[34].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.3487653501372549 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[34].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[34].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[34].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[34].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.69168131558066 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[35].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[35].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221436 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[35].ttft_s", "statistic": "single", "unit": "s", "value": 201.80751334497472 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[35].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.18411419981301616 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[35].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[35].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[35].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[35].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.34378720218977976 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[36].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[36].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241567 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[36].ttft_s", "statistic": "single", "unit": "s", "value": 222.4633266259916 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[36].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.29460413406035335 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[36].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[36].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[36].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[36].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9039437322661442 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[37].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[37].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181179 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[37].ttft_s", "statistic": "single", "unit": "s", "value": 674.9378849080531 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[37].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.1243261503137378 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[37].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[37].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[37].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[37].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7121645716718856 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[38].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[38].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201309 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[38].ttft_s", "statistic": "single", "unit": "s", "value": 157.0683799959952 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[38].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.26611578445020106 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[38].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[38].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360027 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[38].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[38].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7121397237364533 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[39].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[39].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221443 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[39].ttft_s", "statistic": "single", "unit": "s", "value": 171.145450445998 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[39].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.1867505406348614 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[39].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[39].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[39].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[39].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6547196433588469 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[40].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[40].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241573 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[40].ttft_s", "statistic": "single", "unit": "s", "value": 194.45893430901924 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[40].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.29331368349818554 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[40].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[40].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[40].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[40].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7121734162707269 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[41].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[41].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181175 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[41].ttft_s", "statistic": "single", "unit": "s", "value": 585.697342207015 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[41].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.146505949349361 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[41].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[41].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420032 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[41].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[41].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7474227198043636 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[42].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[42].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20118 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[42].ttft_s", "statistic": "single", "unit": "s", "value": 79.51048655901104 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[42].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.823332009651795 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[42].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[42].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[42].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[42].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6579069407945587 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[43].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[43].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40250 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[43].ttft_s", "statistic": "single", "unit": "s", "value": 91.76182534598047 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[43].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.676242059280805 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[43].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[43].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[43].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[43].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6831878430233447 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[44].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[44].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[44].ttft_s", "statistic": "single", "unit": "s", "value": 114.3456150930142 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[44].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.8769080656312824 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[44].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[44].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[44].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[44].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.714141784968022 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[45].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[45].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80509 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[45].ttft_s", "statistic": "single", "unit": "s", "value": 156.6747548970161 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[45].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.45320646501126 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[45].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[45].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[45].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[45].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7722619569655089 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[46].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[46].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100641 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[46].ttft_s", "statistic": "single", "unit": "s", "value": 174.41760934499325 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[46].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.231452506084258 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[46].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[46].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[46].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[46].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8151799370459877 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[47].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[47].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120774 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[47].ttft_s", "statistic": "single", "unit": "s", "value": 190.16879605903523 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[47].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.9782153233635498 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[47].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[47].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120017 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[47].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[47].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8553254149783692 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[48].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[48].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 140909 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[48].ttft_s", "statistic": "single", "unit": "s", "value": 196.07107177498983 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[48].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.315346109947638 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[48].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[48].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140019 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[48].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[48].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8531435776485088 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[49].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[49].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 161039 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[49].ttft_s", "statistic": "single", "unit": "s", "value": 213.34597823995864 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[49].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.674999639183919 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[49].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[49].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160022 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[49].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[49].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8804022329845835 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[50].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[50].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[50].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181172 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[50].ttft_s", "statistic": "single", "unit": "s", "value": 193.97330937400693 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[50].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.37418095717031097 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[50].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[50].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[50].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[50].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7010519256324594 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[51].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[51].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[51].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201304 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[51].ttft_s", "statistic": "single", "unit": "s", "value": 236.95953829103382 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[51].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.6368708136087928 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[51].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[51].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[51].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[51].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8927667520504201 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[52].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[52].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[52].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221432 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[52].ttft_s", "statistic": "single", "unit": "s", "value": 208.1121710090083 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[52].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.37105788422796143 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[52].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[52].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[52].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[52].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7312320684516267 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[53].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[53].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[53].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241560 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[53].ttft_s", "statistic": "single", "unit": "s", "value": 211.3681128610042 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[53].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.7665130845189384 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[53].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[53].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[53].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[53].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9063798703681333 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[54].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[54].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[54].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181165 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[54].ttft_s", "statistic": "single", "unit": "s", "value": 699.4067312839907 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[54].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.18275359890414664 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[54].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[54].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260027 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[54].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[54].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6278306508527068 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[55].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[55].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[55].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201296 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[55].ttft_s", "statistic": "single", "unit": "s", "value": 216.4875847910298 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[55].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.31418965727689213 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[55].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[55].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[55].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[55].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8958130477117819 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[56].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[56].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[56].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221425 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[56].ttft_s", "statistic": "single", "unit": "s", "value": 178.8604112759931 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[56].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.32124962021387304 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[56].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[56].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300029 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[56].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[56].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7048360792790348 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[57].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[57].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[57].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241553 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[57].ttft_s", "statistic": "single", "unit": "s", "value": 170.98452270100825 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[57].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.5603414395285787 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[57].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[57].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320030 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[57].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[57].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9063488312231778 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[58].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[58].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[58].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181151 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[58].ttft_s", "statistic": "single", "unit": "s", "value": 658.1668398780166 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[58].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.18596765244705238 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[58].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[58].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340030 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[58].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[58].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6510017386302731 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[59].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[59].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[59].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201282 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[59].ttft_s", "statistic": "single", "unit": "s", "value": 195.72447070397902 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[59].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.29795097927877384 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[59].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[59].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360030 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[59].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[59].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6432366948484401 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[60].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[60].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[60].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221413 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[60].ttft_s", "statistic": "single", "unit": "s", "value": 123.06476350105368 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[60].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.2727069619654422 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[60].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[60].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380031 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[60].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[60].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6309190470907522 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[61].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[61].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[61].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241546 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[61].ttft_s", "statistic": "single", "unit": "s", "value": 131.52090578101343 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[61].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.66062665313622 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[61].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[61].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400031 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[61].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[61].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9063392811447254 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[62].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[62].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[62].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181157 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[62].ttft_s", "statistic": "single", "unit": "s", "value": 483.53126178798266 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[62].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.412183865338671 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[62].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[62].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420031 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[62].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[62].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6025117150890347 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[63].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[63].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[63].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20119 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[63].ttft_s", "statistic": "single", "unit": "s", "value": 21.97022557305172 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[63].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.655448088466149 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[63].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[63].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[63].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[63].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7011682044957022 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[64].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[64].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[64].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40251 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[64].ttft_s", "statistic": "single", "unit": "s", "value": 34.32781493803486 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[64].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.139958296986551 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[64].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[64].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[64].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[64].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7613421048471422 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[65].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[65].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[65].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60385 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[65].ttft_s", "statistic": "single", "unit": "s", "value": 132.55791806901107 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[65].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.87786222054429 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[65].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[65].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60005 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[65].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[65].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6491678396952886 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[66].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[66].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[66].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80512 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[66].ttft_s", "statistic": "single", "unit": "s", "value": 173.90474537096452 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[66].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.233873909176997 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[66].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[66].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80005 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[66].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[66].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8151799370459877 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[67].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[67].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[67].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100645 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[67].ttft_s", "statistic": "single", "unit": "s", "value": 190.15485956601333 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[67].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.774577196786765 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[67].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[67].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100008 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[67].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[67].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8427175871771364 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[68].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[68].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[68].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120773 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[68].ttft_s", "statistic": "single", "unit": "s", "value": 194.91742883599363 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[68].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.4000918780454636 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[68].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[68].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[68].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[68].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8650683874881357 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[69].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[69].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[69].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 140904 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[69].ttft_s", "statistic": "single", "unit": "s", "value": 212.42995198100107 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[69].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.7082827449885807 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[69].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[69].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[69].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[69].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8804022329845835 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[70].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[70].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[70].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 161039 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[70].ttft_s", "statistic": "single", "unit": "s", "value": 201.03466660296544 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[70].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.3796269792896577 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[70].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[70].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[70].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[70].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7010519256324594 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[71].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[71].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[71].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181170 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[71].ttft_s", "statistic": "single", "unit": "s", "value": 237.34325001301477 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[71].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.7374118042455753 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[71].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[71].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180016 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[71].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[71].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8927667520504201 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[72].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[72].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[72].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201307 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[72].ttft_s", "statistic": "single", "unit": "s", "value": 213.38447414099937 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[72].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.28634270451185023 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[72].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[72].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200018 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[72].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[72].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8017224353995854 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[73].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[73].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[73].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221443 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[73].ttft_s", "statistic": "single", "unit": "s", "value": 191.56268116104184 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[73].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.194305985006265 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[73].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[73].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220018 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[73].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[73].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9063461026087978 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[74].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[74].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[74].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241572 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[74].ttft_s", "statistic": "single", "unit": "s", "value": 229.2531233089976 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[74].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.31634115684589126 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[74].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[74].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240018 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[74].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[74].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7191422455173179 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[75].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[75].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[75].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181185 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[75].ttft_s", "statistic": "single", "unit": "s", "value": 873.9897076139459 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[75].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.24722537976393036 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[75].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[75].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[75].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[75].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.555262070991724 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[76].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[76].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[76].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201318 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[76].ttft_s", "statistic": "single", "unit": "s", "value": 116.80983015702805 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[76].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.22672302729788332 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[76].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[76].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[76].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[76].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9039113593120204 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[77].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[77].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[77].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221454 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[77].ttft_s", "statistic": "single", "unit": "s", "value": 109.40606641402701 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[77].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.5037453348143096 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[77].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[77].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[77].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[77].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7524648167848422 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[78].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[78].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[78].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241589 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[78].ttft_s", "statistic": "single", "unit": "s", "value": 124.11030960595235 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[78].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.23780927585904432 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[78].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[78].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[78].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[78].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7257874231317646 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[79].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[79].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[79].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181195 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[79].ttft_s", "statistic": "single", "unit": "s", "value": 771.7500437250128 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[79].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.25196968965367056 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[79].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[79].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[79].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[79].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.49850847844931434 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[80].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[80].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[80].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201331 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[80].ttft_s", "statistic": "single", "unit": "s", "value": 90.80128397396766 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[80].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.24398933368395284 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[80].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[80].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[80].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[80].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9030281109660554 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[81].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[81].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[81].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221468 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[81].ttft_s", "statistic": "single", "unit": "s", "value": 56.389698530954774 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[81].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.4211015402595004 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[81].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[81].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380025 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[81].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[81].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.64319958979618 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[82].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[82].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[82].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241604 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[82].ttft_s", "statistic": "single", "unit": "s", "value": 106.95489919296233 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[82].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.29321310222388874 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[82].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[82].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400026 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[82].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[82].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.915122064289676 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[83].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[83].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[83].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181200 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[83].ttft_s", "statistic": "single", "unit": "s", "value": 398.9729665360064 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[83].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.34239520642968 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[83].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[83].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420027 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[83].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[83].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[84].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[84].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[84].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20118 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[84].ttft_s", "statistic": "single", "unit": "s", "value": 95.44786189403385 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[84].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.7106911852666804 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[84].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[84].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[84].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[84].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.3116831486358893 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[85].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[85].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[85].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40249 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[85].ttft_s", "statistic": "single", "unit": "s", "value": 155.08917450899025 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[85].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.431367844316341 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[85].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[85].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[85].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[85].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7722619569655089 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[86].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[86].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[86].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60380 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[86].ttft_s", "statistic": "single", "unit": "s", "value": 170.42058520804858 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[86].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.69459686572708 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[86].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[86].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[86].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[86].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8151799370459877 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[87].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[87].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[87].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 80514 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[87].ttft_s", "statistic": "single", "unit": "s", "value": 190.2149516909849 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[87].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.869670828947707 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[87].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[87].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[87].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[87].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8427175871771364 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[88].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[88].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[88].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 100647 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[88].ttft_s", "statistic": "single", "unit": "s", "value": 202.228887210018 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[88].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.355514718696939 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[88].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[88].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[88].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[88].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8650683874881357 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[89].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[89].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[89].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 120782 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[89].ttft_s", "statistic": "single", "unit": "s", "value": 218.10660801897757 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[89].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.143220035975215 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[89].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[89].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[89].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[89].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8804022329845835 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[90].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[90].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[90].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 140917 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[90].ttft_s", "statistic": "single", "unit": "s", "value": 197.372952411999 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[90].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.3806748561777727 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[90].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[90].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[90].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[90].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7010519256324594 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[91].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[91].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[91].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 161049 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[91].ttft_s", "statistic": "single", "unit": "s", "value": 231.76345212798333 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[91].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.4545089366570143 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[91].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[91].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160017 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[91].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[91].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.8927667520504201 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[92].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[92].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[92].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181185 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[92].ttft_s", "statistic": "single", "unit": "s", "value": 213.42413677903824 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[92].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.33483327682121666 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[92].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[92].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[92].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[92].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.788668724113735 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[93].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[93].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[93].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201317 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[93].ttft_s", "statistic": "single", "unit": "s", "value": 221.37890139903175 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[93].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.6461478143525987 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[93].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[93].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[93].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[93].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9012773204655123 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[94].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[94].turn_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[94].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221449 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[94].ttft_s", "statistic": "single", "unit": "s", "value": 229.27614972699666 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[94].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.3603659341088923 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[94].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[94].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 220020 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[94].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[94].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6491084615135929 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[95].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[95].turn_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[95].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241581 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[95].ttft_s", "statistic": "single", "unit": "s", "value": 198.5167554700165 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[95].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.18788464040724603 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[95].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[95].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 240021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[95].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[95].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.654751149484168 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[96].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[96].turn_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[96].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181182 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[96].ttft_s", "statistic": "single", "unit": "s", "value": 709.0973409049911 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[96].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.2558251567365378 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[96].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[96].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 260021 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[96].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[96].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7317019549482514 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[97].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[97].turn_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[97].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201320 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[97].ttft_s", "statistic": "single", "unit": "s", "value": 157.91423595201923 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[97].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.661084170481826 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[97].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[97].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 280023 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[97].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[97].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7474669891801318 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[98].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[98].turn_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[98].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221457 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[98].ttft_s", "statistic": "single", "unit": "s", "value": 158.76458215998719 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[98].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.2599740903626114 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[98].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[98].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 300024 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[98].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[98].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7191239685270051 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[99].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[99].turn_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[99].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241595 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[99].ttft_s", "statistic": "single", "unit": "s", "value": 163.11555990198394 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[99].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.182013214008923 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[99].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20004 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[99].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 320028 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[99].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[99].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.7191123381986843 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[100].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[100].turn_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[100].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181189 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[100].ttft_s", "statistic": "single", "unit": "s", "value": 604.8709072089987 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[100].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.2709420898646213 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[100].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[100].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 340031 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[100].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[100].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5864799630644794 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[101].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[101].turn_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[101].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 201320 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[101].ttft_s", "statistic": "single", "unit": "s", "value": 127.62333272595424 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[101].decode_tps", "statistic": "single", "unit": "tok/s", "value": 1.2209389929085264 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[101].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[101].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 360033 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[101].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[101].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.9012176315316424 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[102].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[102].turn_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[102].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 221455 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[102].ttft_s", "statistic": "single", "unit": "s", "value": 104.37620295304805 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[102].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.3293056313095077 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[102].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[102].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 380034 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[102].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[102].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.6547369205359693 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[103].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[103].turn_number", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[103].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 241584 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[103].ttft_s", "statistic": "single", "unit": "s", "value": 111.28352323401487 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[103].decode_tps", "statistic": "single", "unit": "tok/s", "value": 0.38746386745055705 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[103].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[103].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 400034 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[103].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[103].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5229336966394187 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[104].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[104].turn_number", "statistic": "single", "unit": "count", "value": 21 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[104].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 181188 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[104].ttft_s", "statistic": "single", "unit": "s", "value": 414.85925020003924 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[104].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.862749670149645 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[104].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[104].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 420034 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[104].dropped_messages", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[104].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 675.7934919499676 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 237.34325001301477 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 173.58133394946344 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 105 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "60/105" } ], "reasons": [ "60/105 requests passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "d4b1b1ba698117f3d6006afdedd664b9", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Measured retrieval only, with no customer qualification threshold. Exact dense float32 array size excludes payloads and vector database overhead. One timing run; no reranking, ingestion, answer generation or GPU measurement.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Transformers CPU", "recipe": "plain", "version": "5.19.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "c937279f23ea467192959b8e", "measured_at": "2026-10-07T14:39:45.951891+00:00", "metrics": [ { "basis": null, "n": null, "name": "passages_per_second", "scope": "machine_total", "source_field": "passages_per_second", "statistic": "single", "unit": "passages/s", "value": 12.25921414124026 }, { "basis": null, "n": null, "name": "query_seconds", "scope": "per_person", "source_field": "query_seconds", "statistic": "single", "unit": "s", "value": 2.574559000087902 }, { "basis": null, "n": null, "name": "recall_at_5", "scope": "machine_total", "source_field": "recall_at_5", "statistic": "single", "unit": "ratio", "value": 1.0 }, { "basis": null, "n": null, "name": "mrr", "scope": "machine_total", "source_field": "mrr", "statistic": "single", "unit": "ratio", "value": 0.9133333333333334 }, { "basis": null, "n": null, "name": "questions", "scope": "machine_total", "source_field": "questions", "statistic": "single", "unit": "count", "value": 100 }, { "basis": null, "n": null, "name": "passage_seconds", "scope": "machine_total", "source_field": "passage_seconds", "statistic": "single", "unit": "s", "value": 32.62851887498982 }, { "basis": null, "n": null, "name": "vector_bytes", "scope": "machine_total", "source_field": "vector_bytes", "statistic": "single", "unit": "bytes", "value": 204800 }, { "basis": null, "n": null, "name": "passages", "scope": "machine_total", "source_field": "passages", "statistic": "single", "unit": "count", "value": 400 }, { "basis": null, "n": null, "name": "index_file_bytes", "scope": "machine_total", "source_field": "index_file_bytes", "statistic": "single", "unit": "bytes", "value": 204928 }, { "basis": null, "n": null, "name": "recall_at_10", "scope": "machine_total", "source_field": "recall_at_10", "statistic": "single", "unit": "ratio", "value": 1.0 }, { "basis": null, "n": null, "name": "truncated_passages", "scope": "machine_total", "source_field": "truncated_passages", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "google/embeddinggemma-2", "publisher_revision": "914f7f89142e33e77833254d9c9b90c3cef7303b", "variant": "FP32 variant", "weights_dtype": "FP32", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-embedding-eval-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "embeddings", "version": "1" }, "settings": { "context_tokens": 262144, "corpus_sha256": "d7fb630b25a7d5128c44491d1a8f21bde479711f79520f1ce95ea8bd8a0ae1f6", "dimensions": 128, "dtype": "float32" }, "setup_complete": false, "setup_id": "a1cab9aac7b0159473194b220292b58c", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple Silicon laptop CPU, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "d0edaadb8a1940038d9fd76e", "measured_at": "2026-10-07T07:22:01.496596+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 18.157898496021517 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 14.437703111965675 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 14.450682270049583 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 14.450402253016364 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 17.44542896101484 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 36.05977860283406 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 109 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[0].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 14.429424229019787 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 10.89551008300623 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 17.683375458000228 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 17.973249800952022 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 123 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[1].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 10.892245573049877 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 13.266313811996952 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 17.72674516902771 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 27.127421165056983 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 122 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[2].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 13.259319078992121 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 14.450682270049583 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 17.95352709305007 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 37.11269170315226 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 131 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[3].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 14.437643639044836 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 14.437703111965675 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 18.135093041986693 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 37.323626290942904 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 139 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[4].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 14.432861714973114 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 85.36082827820009 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 34.365210276768615 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 94.23422622680664 } ], "model": { "artifact_sha256": [ "648fe83171485f4be03ea58d86089e088c94c2a3dc88922233f9002f30d8bc8b", "8303cef7d375021d69dcafc60d3fad5f8186995b18b0db5e2a609aadeb8a4f51", "cb5c5978721a1e87406ae68f1cd5a855326e40ff8bc13832644c392115336987" ], "publisher_repo": "ornith-ai/Ornith-1.5-35B-A3B", "publisher_revision": "10fbf86fed7ecee4a061f8b499a618f46001cac1", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": true, "name": "first-token-300-s-reference", "observed": 14.450682270049583 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "77fda79ee1961c35a8f8a77e7d5a01d5f9ce414975e00e640a166e71e5f9093e", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 36.05977860283406, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 14.429424229019787, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 109, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 17.44542896101484, "ttft_s": 14.450402253016364, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 17.973249800952022, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 10.892245573049877, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 123, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 17.683375458000228, "ttft_s": 10.89551008300623, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 27.127421165056983, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 13.259319078992121, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 122, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 17.72674516902771, "ttft_s": 13.266313811996952, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 37.11269170315226, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 14.437643639044836, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 131, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 17.95352709305007, "ttft_s": 14.450682270049583, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 37.323626290942904, "effective_prefill_tps": null, "first_real_token_from_dispatch_s": 14.432861714973114, "first_token_within_300_seconds": true, "needle_correct": 3, "needle_total": 3, "output_tokens": 139, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 18.135093041986693, "ttft_s": 14.437703111965675, "user_number": 2, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "warm", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "shared-record-groups-v2", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "shared", "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "bde5d43ff70430becda39144f69c34e0", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": 250000, "document": "shared", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Bounded fictional tasks only. Unsupported cases are separately counted, never passes. Original publisher name with variant; exact variant repository and revision retained privately. Not customer acceptance.", "Publisher revision was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "d490228d2adf4b9fa03f0331", "measured_at": "2026-10-07T04:23:11.246879+00:00", "metrics": [ { "basis": null, "n": null, "name": "skipped_cases", "scope": "machine_total", "source_field": "skipped_cases", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "correctness_failures", "scope": "machine_total", "source_field": "correctness_failures", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 0.057674845040310174 }, { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 130 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 102 }, { "basis": null, "n": null, "name": "api_cases_passed", "scope": "machine_total", "source_field": "api_cases_passed", "statistic": "single", "unit": "count", "value": 100 }, { "basis": null, "n": null, "name": "ttft_p95_s", "scope": "per_person", "source_field": "ttft_p95_s", "statistic": "p95", "unit": "s", "value": 0.12166704901028425 }, { "basis": null, "n": null, "name": "completion_p95_s", "scope": "per_person", "source_field": "completion_p95_s", "statistic": "p95", "unit": "s", "value": 0.8050042009563185 }, { "basis": null, "n": null, "name": "api_cases_total", "scope": "machine_total", "source_field": "api_cases_total", "statistic": "single", "unit": "count", "value": 124 }, { "basis": null, "n": null, "name": "completion_median_s", "scope": "per_person", "source_field": "completion_median_s", "statistic": "median", "unit": "s", "value": 0.19479476602282375 }, { "basis": null, "n": null, "name": "completeness_failures", "scope": "machine_total", "source_field": "completeness_failures", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[0].passed", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[0].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[0].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[1].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[1].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[1].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[2].passed", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[2].total", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[2].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[3].passed", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[3].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[3].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[4].passed", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[4].total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[4].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[5].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[5].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[5].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[6].passed", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[6].total", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[6].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[7].passed", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[7].total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[7].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[8].passed", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[8].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[8].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[9].passed", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[9].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[9].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[10].passed", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[10].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[10].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[11].passed", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[11].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[11].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[12].passed", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[12].total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[12].skipped", "statistic": "single", "unit": "count", "value": 4 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "structured-tasks", "observed": "102/130" }, { "met": false, "name": "API-cases", "observed": "100/124" } ], "reasons": [ "102/130 tasks passed.", "100/124 API cases passed." ], "status": "partial" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 102, "skipped": null, "suite": "structured-tasks", "total": 130, "version": "core-v1" }, { "passed": 100, "skipped": 4, "suite": "API-cases", "total": 124, "version": "core-v1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "42c4dc34a066f791a62c9b1dbeb97943988c746309a43a464f0417a02ae9f1da", "harness": { "name": "keep-api-suite-run-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": "core-v1", "workload_version": "1" }, "requests": [], "scenario": { "id": "api-suite", "version": "1" }, "settings": { "concurrency": 1, "context_tokens": 262144, "kv_dtype": "fp8", "speculative_decoding": "plain" }, "setup_complete": false, "setup_id": "b7217e9367ab44e63a8bd94b11ecf8d3", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Finite lab cohort; configured 262144 context differs from measured inputs. Aggregate uses first-output to final-completion window. Whole-host memory sampled during speed only; missing counters remain unassessed. Broader coding quality and delivery acceptance unassessed." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "d65163d3e20a4aac92032dc2", "measured_at": "2026-10-07T04:19:07.006640+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 202.04653906822205 }, { "basis": null, "n": 20, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 114.3505926100188 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 60 }, { "basis": null, "n": 20, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 162.46046369097894 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 60 }, { "basis": null, "n": null, "name": "preemptions", "scope": "machine_total", "source_field": "preemptions", "statistic": "single", "unit": "count", "value": 0.0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "correctness_probes_passed", "scope": "machine_total", "source_field": "correctness_probes_passed", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 71.51700109014168 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 50.68139274854103 }, { "basis": null, "n": null, "name": "correctness_probes_total", "scope": "machine_total", "source_field": "correctness_probes_total", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": 20, "name": "decode_tps", "scope": "per_person", "source_field": "per_request_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 6.893528752744044 }, { "basis": null, "n": null, "name": "forced_output_tokens", "scope": "machine_total", "source_field": "forced_output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 93.79351043701172 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 93.53175729000941 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[0].total_s", "statistic": "single", "unit": "s", "value": 199.94406931602862 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.802075908989307 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40100 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 162.12879321700893 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[1].total_s", "statistic": "single", "unit": "s", "value": 202.01941937801894 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.810027045889365 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 154.09069847100182 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[2].total_s", "statistic": "single", "unit": "s", "value": 201.7493742229999 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.72207718609505 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[2].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 125.18151886499254 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[3].total_s", "statistic": "single", "unit": "s", "value": 200.92618343001232 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.746349765155457 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40100 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 91.9306046019774 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[4].total_s", "statistic": "single", "unit": "s", "value": 199.8824033520068 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.733594121791887 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[5].ttft_s", "statistic": "single", "unit": "s", "value": 93.53399686899502 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[5].total_s", "statistic": "single", "unit": "s", "value": 199.94388840498868 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.802185141098013 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[5].user_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40101 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[5].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[5].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[5].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[6].ttft_s", "statistic": "single", "unit": "s", "value": 62.44563609204488 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[6].total_s", "statistic": "single", "unit": "s", "value": 198.7315115420497 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.7494714570583323 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[6].user_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40106 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[6].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[6].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[6].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[7].ttft_s", "statistic": "single", "unit": "s", "value": 60.890863093023654 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[7].total_s", "statistic": "single", "unit": "s", "value": 198.65202202403452 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.7093183881815532 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[7].user_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40106 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[7].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[7].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[7].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[8].ttft_s", "statistic": "single", "unit": "s", "value": 108.32101823401172 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[8].total_s", "statistic": "single", "unit": "s", "value": 200.42742160096532 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.547931319868898 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[8].user_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[8].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[8].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[8].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[9].ttft_s", "statistic": "single", "unit": "s", "value": 62.44542540400289 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[9].total_s", "statistic": "single", "unit": "s", "value": 198.73236480297055 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.749442186122427 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[9].user_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[9].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[9].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[9].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[10].ttft_s", "statistic": "single", "unit": "s", "value": 120.38016698602587 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[10].total_s", "statistic": "single", "unit": "s", "value": 200.78525526000885 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.355319184014209 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[10].user_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[10].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[10].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[10].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[11].ttft_s", "statistic": "single", "unit": "s", "value": 62.44689436000772 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[11].total_s", "statistic": "single", "unit": "s", "value": 198.72969399398426 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.7495560802421544 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[11].user_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40099 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[11].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[11].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[11].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[12].ttft_s", "statistic": "single", "unit": "s", "value": 162.46046369097894 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[12].total_s", "statistic": "single", "unit": "s", "value": 202.04564789996948 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.908870078819598 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[12].user_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[12].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[12].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[12].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[13].ttft_s", "statistic": "single", "unit": "s", "value": 146.2711280939984 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[13].total_s", "statistic": "single", "unit": "s", "value": 201.5089382849983 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.250909806762385 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[13].user_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40099 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[13].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[13].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[13].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[14].ttft_s", "statistic": "single", "unit": "s", "value": 126.69506250100676 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[14].total_s", "statistic": "single", "unit": "s", "value": 200.97121076699113 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.879732079941716 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[14].user_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40100 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[14].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[14].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[14].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[15].ttft_s", "statistic": "single", "unit": "s", "value": 138.4330267990008 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[15].total_s", "statistic": "single", "unit": "s", "value": 201.29494035604876 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 8.128928489207718 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[15].user_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[15].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[15].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[15].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[16].ttft_s", "statistic": "single", "unit": "s", "value": 159.39992942800745 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[16].total_s", "statistic": "single", "unit": "s", "value": 201.91818897501798 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.018365884309311 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[16].user_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[16].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[16].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[16].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[17].ttft_s", "statistic": "single", "unit": "s", "value": 149.4206585349748 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[17].total_s", "statistic": "single", "unit": "s", "value": 201.60680380096892 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.791870953399984 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[17].user_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[17].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[17].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[17].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[18].ttft_s", "statistic": "single", "unit": "s", "value": 59.14211418799823 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[18].total_s", "statistic": "single", "unit": "s", "value": 198.56782370398287 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.66503424493182 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[18].user_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40099 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[18].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[18].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[18].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[19].ttft_s", "statistic": "single", "unit": "s", "value": 62.44722560798982 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[19].total_s", "statistic": "single", "unit": "s", "value": 198.7314917339827 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.7495157330016933 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[19].user_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40101 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[19].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[19].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[19].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 20, "name": "decode_tps", "scope": "per_person", "source_field": "per_person_decode_median_tps", "statistic": "median", "unit": "tok/s", "value": 5.951625251941554 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "20/20" }, { "met": true, "name": "hidden-facts", "observed": "60/60" }, { "met": true, "name": "correctness-probes", "observed": "20/20" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 60, "skipped": null, "suite": "hidden-facts", "total": 60, "version": "1" }, { "passed": 20, "skipped": null, "suite": "correctness-probes", "total": 20, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "e2097959b1ca0142f2b93edc3ec411e474c0a6622800dd247af7a71bd9c39e83", "harness": { "name": "keep-team-load-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 4.802075908989307, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40100, "total_s": 199.94406931602862, "ttft_s": 93.53175729000941, "user_number": 0 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 12.810027045889365, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 202.01941937801894, "ttft_s": 162.12879321700893, "user_number": 1 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 10.72207718609505, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 201.7493742229999, "ttft_s": 154.09069847100182, "user_number": 2 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 6.746349765155457, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40100, "total_s": 200.92618343001232, "ttft_s": 125.18151886499254, "user_number": 3 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 4.733594121791887, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 199.8824033520068, "ttft_s": 91.9306046019774, "user_number": 4 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 4.802185141098013, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40101, "total_s": 199.94388840498868, "ttft_s": 93.53399686899502, "user_number": 5 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.7494714570583323, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40106, "total_s": 198.7315115420497, "ttft_s": 62.44563609204488, "user_number": 6 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.7093183881815532, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40106, "total_s": 198.65202202403452, "ttft_s": 60.890863093023654, "user_number": 7 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.547931319868898, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 200.42742160096532, "ttft_s": 108.32101823401172, "user_number": 8 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.749442186122427, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 198.73236480297055, "ttft_s": 62.44542540400289, "user_number": 9 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 6.355319184014209, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 200.78525526000885, "ttft_s": 120.38016698602587, "user_number": 10 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.7495560802421544, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40099, "total_s": 198.72969399398426, "ttft_s": 62.44689436000772, "user_number": 11 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 12.908870078819598, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 202.04564789996948, "ttft_s": 162.46046369097894, "user_number": 12 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 9.250909806762385, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40099, "total_s": 201.5089382849983, "ttft_s": 146.2711280939984, "user_number": 13 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 6.879732079941716, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40100, "total_s": 200.97121076699113, "ttft_s": 126.69506250100676, "user_number": 14 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 8.128928489207718, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 201.29494035604876, "ttft_s": 138.4330267990008, "user_number": 15 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 12.018365884309311, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 201.91818897501798, "ttft_s": 159.39992942800745, "user_number": 16 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 9.791870953399984, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 201.60680380096892, "ttft_s": 149.4206585349748, "user_number": 17 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.66503424493182, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40099, "total_s": 198.56782370398287, "ttft_s": 59.14211418799823, "user_number": 18 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.7495157330016933, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40101, "total_s": 198.7314917339827, "ttft_s": 62.44722560798982, "user_number": 19 } ], "scenario": { "id": "team-load", "version": "1" }, "settings": { "concurrency": 20, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "plain", "thinking": false }, "setup_complete": false, "setup_id": "a4b96c03f67c006a12a64bbc7734ee71", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": { "max": 40106, "median": 40102.0, "min": 40099 }, "output_tokens": 512, "repetitions": null, "users": 20, "warmup": null } }, { "caveats": [ "Bounded fictional tasks only. Unsupported cases are separately counted, never passes. Original publisher name with variant; exact variant repository and revision retained privately. Not customer acceptance.", "Publisher revision was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "dae6c36fbef44bb583b17235", "measured_at": "2026-10-07T04:19:53.741073+00:00", "metrics": [ { "basis": null, "n": null, "name": "skipped_cases", "scope": "machine_total", "source_field": "skipped_cases", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "correctness_failures", "scope": "machine_total", "source_field": "correctness_failures", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "completeness_failures", "scope": "machine_total", "source_field": "completeness_failures", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "structured-tasks", "observed": "0/0" } ], "reasons": [ "Bounded fictional tasks only." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "structured-tasks", "total": 0, "version": "core-v1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "42c4dc34a066f791a62c9b1dbeb97943988c746309a43a464f0417a02ae9f1da", "harness": { "name": "keep-api-suite-run-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": "core-v1", "workload_version": "1" }, "requests": [], "scenario": { "id": "api-suite", "version": "1" }, "settings": { "concurrency": 1, "context_tokens": 262144, "kv_dtype": "fp8", "speculative_decoding": "plain" }, "setup_complete": false, "setup_id": "b7217e9367ab44e63a8bd94b11ecf8d3", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Startup, inference completion, concurrent decode and retrieval correctness are separate gates.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "e01024f25d5242c2b407f991", "measured_at": "2026-10-07T20:14:25.383439+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "INT4 variant", "weights_dtype": "INT4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/0" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "did_not_start" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "long-context-concurrency", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "2cdb4314c031e68844e1f2572dc8372d", "standard": "pre-standard", "system": { "driver": null, "hardware": "unrecorded hardware", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Finite lab cohort; configured 262144 context differs from measured inputs. Aggregate uses first-output to final-completion window. Whole-host memory sampled during speed only; missing counters remain unassessed. Broader coding quality and delivery acceptance unassessed." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "e02e754e7dba478e91d3f202", "measured_at": "2026-10-07T04:42:29.268039+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 155.528804063797 }, { "basis": null, "n": 16, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 92.14591929697781 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 48 }, { "basis": null, "n": 16, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 132.4827625619946 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 48 }, { "basis": null, "n": null, "name": "preemptions", "scope": "machine_total", "source_field": "preemptions", "statistic": "single", "unit": "count", "value": 0.0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "mtp_acceptance_rate", "scope": "machine_total", "source_field": "mtp_acceptance_rate", "statistic": "single", "unit": "ratio", "value": 0.847737556561086 }, { "basis": null, "n": null, "name": "correctness_probes_passed", "scope": "machine_total", "source_field": "correctness_probes_passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 83.08403211684993 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 52.671915336272306 }, { "basis": null, "n": null, "name": "correctness_probes_total", "scope": "machine_total", "source_field": "correctness_probes_total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": 16, "name": "decode_tps", "scope": "per_person", "source_field": "per_request_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 10.93416521174156 }, { "basis": null, "n": null, "name": "forced_output_tokens", "scope": "machine_total", "source_field": "forced_output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 93.619873046875 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 60.670777673018165 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[0].total_s", "statistic": "single", "unit": "s", "value": 152.4296711490024 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.568942482221001 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40101 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 90.55281917803222 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[1].total_s", "statistic": "single", "unit": "s", "value": 153.7877831220394 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 8.080972426148238 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 120.40713858703384 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[2].total_s", "statistic": "single", "unit": "s", "value": 154.79961349302903 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 14.857901369317394 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[2].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 59.03314933698857 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[3].total_s", "statistic": "single", "unit": "s", "value": 152.00303491699742 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.4964034516342295 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40107 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 60.670666008023545 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[4].total_s", "statistic": "single", "unit": "s", "value": 152.64704256498953 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.555774418700978 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[5].ttft_s", "statistic": "single", "unit": "s", "value": 127.96538553497521 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[5].total_s", "statistic": "single", "unit": "s", "value": 155.22761317697586 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 18.743882807755032 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[5].user_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40101 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[5].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[5].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[5].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[6].ttft_s", "statistic": "single", "unit": "s", "value": 132.4827625619946 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[6].total_s", "statistic": "single", "unit": "s", "value": 155.52602438797476 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 22.1756799822442 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[6].user_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[6].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[6].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[6].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[7].ttft_s", "statistic": "single", "unit": "s", "value": 131.56556399998954 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[7].total_s", "statistic": "single", "unit": "s", "value": 155.22646096698008 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 21.596814385900036 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[7].user_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40106 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[7].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[7].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[7].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[8].ttft_s", "statistic": "single", "unit": "s", "value": 119.4638219680055 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[8].total_s", "statistic": "single", "unit": "s", "value": 154.84495083097136 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 14.44272742057346 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[8].user_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[8].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[8].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[8].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[9].ttft_s", "statistic": "single", "unit": "s", "value": 120.4056167450035 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[9].total_s", "statistic": "single", "unit": "s", "value": 154.92426558904117 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 14.803592177341667 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[9].user_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40101 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[9].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[9].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[9].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[10].ttft_s", "statistic": "single", "unit": "s", "value": 60.67007112602005 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[10].total_s", "statistic": "single", "unit": "s", "value": 152.0937898660195 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.589359162398944 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[10].user_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40100 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[10].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[10].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[10].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[11].ttft_s", "statistic": "single", "unit": "s", "value": 92.14634043298429 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[11].total_s", "statistic": "single", "unit": "s", "value": 154.03607036801986 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 8.256620291224191 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[11].user_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[11].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[11].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[11].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[12].ttft_s", "statistic": "single", "unit": "s", "value": 106.12417342397384 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[12].total_s", "statistic": "single", "unit": "s", "value": 154.46316338097677 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.571176610320771 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[12].user_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[12].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[12].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[12].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[13].ttft_s", "statistic": "single", "unit": "s", "value": 60.668586324027274 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[13].total_s", "statistic": "single", "unit": "s", "value": 152.57779390003998 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.559834683346409 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[13].user_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[13].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[13].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[13].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[14].ttft_s", "statistic": "single", "unit": "s", "value": 92.14549816097133 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[14].total_s", "statistic": "single", "unit": "s", "value": 153.85567999299383 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 8.280643239570445 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[14].user_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[14].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[14].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[14].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[15].ttft_s", "statistic": "single", "unit": "s", "value": 57.12241008004639 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[15].total_s", "statistic": "single", "unit": "s", "value": 152.34597945702262 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.366318479167962 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[15].user_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40106 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[15].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[15].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[15].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 16, "name": "decode_tps", "scope": "per_person", "source_field": "per_person_decode_median_tps", "statistic": "median", "unit": "tok/s", "value": 8.268631765397318 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "16/16" }, { "met": true, "name": "hidden-facts", "observed": "48/48" }, { "met": true, "name": "correctness-probes", "observed": "16/16" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 48, "skipped": null, "suite": "hidden-facts", "total": 48, "version": "1" }, { "passed": 16, "skipped": null, "suite": "correctness-probes", "total": 16, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "e2097959b1ca0142f2b93edc3ec411e474c0a6622800dd247af7a71bd9c39e83", "harness": { "name": "keep-team-load-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.568942482221001, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40101, "total_s": 152.4296711490024, "ttft_s": 60.670777673018165, "user_number": 0 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 8.080972426148238, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 153.7877831220394, "ttft_s": 90.55281917803222, "user_number": 1 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 14.857901369317394, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 154.79961349302903, "ttft_s": 120.40713858703384, "user_number": 2 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.4964034516342295, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40107, "total_s": 152.00303491699742, "ttft_s": 59.03314933698857, "user_number": 3 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.555774418700978, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 152.64704256498953, "ttft_s": 60.670666008023545, "user_number": 4 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 18.743882807755032, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40101, "total_s": 155.22761317697586, "ttft_s": 127.96538553497521, "user_number": 5 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 22.1756799822442, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 155.52602438797476, "ttft_s": 132.4827625619946, "user_number": 6 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 21.596814385900036, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40106, "total_s": 155.22646096698008, "ttft_s": 131.56556399998954, "user_number": 7 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 14.44272742057346, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 154.84495083097136, "ttft_s": 119.4638219680055, "user_number": 8 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 14.803592177341667, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40101, "total_s": 154.92426558904117, "ttft_s": 120.4056167450035, "user_number": 9 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.589359162398944, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40100, "total_s": 152.0937898660195, "ttft_s": 60.67007112602005, "user_number": 10 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 8.256620291224191, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 154.03607036801986, "ttft_s": 92.14634043298429, "user_number": 11 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 10.571176610320771, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 154.46316338097677, "ttft_s": 106.12417342397384, "user_number": 12 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.559834683346409, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 152.57779390003998, "ttft_s": 60.668586324027274, "user_number": 13 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 8.280643239570445, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 153.85567999299383, "ttft_s": 92.14549816097133, "user_number": 14 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 5.366318479167962, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40106, "total_s": 152.34597945702262, "ttft_s": 57.12241008004639, "user_number": 15 } ], "scenario": { "id": "team-load", "version": "1" }, "settings": { "concurrency": 16, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "cf0bbc6439f3c0ed87c21751d34d5655", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": { "max": 40107, "median": 40103.0, "min": 40100 }, "output_tokens": 512, "repetitions": null, "users": 16, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "e1c73db9297c42f0b08b2e20", "measured_at": "2026-10-07T17:10:05.103859+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 4.2007552070426755 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20195 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 3.958434721978847 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.04116448387101 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[0].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20763 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 4.162072690029163 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.54169700649555 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40000 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[1].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20764 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 4.1640529439901 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.82796060913925 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60001 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[2].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20766 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 4.157310820999555 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.05569044822327 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[3].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[4].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20762 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[4].ttft_s", "statistic": "single", "unit": "s", "value": 4.165260132984258 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.8985446793404 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[4].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[4].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[4].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[4].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[5].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[5].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20765 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[5].ttft_s", "statistic": "single", "unit": "s", "value": 4.164349503989797 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.13545439166964 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[5].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[5].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[5].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[5].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[6].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[6].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20763 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[6].ttft_s", "statistic": "single", "unit": "s", "value": 4.164988242031541 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.91008875057591 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[6].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[6].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[6].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[6].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[7].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[7].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20762 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[7].ttft_s", "statistic": "single", "unit": "s", "value": 4.181012828019448 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.72233645720881 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[7].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[7].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[7].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[7].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[8].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[8].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20763 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[8].ttft_s", "statistic": "single", "unit": "s", "value": 4.182111294998322 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.83691903034119 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[8].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[8].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[8].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[8].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[9].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[9].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20767 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[9].ttft_s", "statistic": "single", "unit": "s", "value": 4.199595564976335 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 72.03129874076429 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[9].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[9].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[9].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[9].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[10].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[10].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20201 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[10].ttft_s", "statistic": "single", "unit": "s", "value": 3.9647572839749046 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.9626314688174 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[10].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[10].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[10].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[10].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[11].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[11].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20753 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[11].ttft_s", "statistic": "single", "unit": "s", "value": 4.186333785997704 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.79376961586749 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[11].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[11].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[11].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[11].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[12].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[12].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20753 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[12].ttft_s", "statistic": "single", "unit": "s", "value": 4.1741112260497175 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.62647244990436 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[12].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[12].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60002 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[12].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[12].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[13].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[13].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20756 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[13].ttft_s", "statistic": "single", "unit": "s", "value": 4.17686991795199 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.6086736862249 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[13].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[13].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80004 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[13].dropped_messages", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[13].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[14].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[14].turn_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20757 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[14].ttft_s", "statistic": "single", "unit": "s", "value": 4.1770055330125615 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.31231452662743 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[14].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[14].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 100007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[14].dropped_messages", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[14].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[15].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[15].turn_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20753 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[15].ttft_s", "statistic": "single", "unit": "s", "value": 4.1750841560424305 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.50461339382697 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[15].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20000 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[15].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 120007 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[15].dropped_messages", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[15].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[16].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[16].turn_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20757 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[16].ttft_s", "statistic": "single", "unit": "s", "value": 4.180532122962177 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.75668492738237 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[16].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[16].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 140010 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[16].dropped_messages", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[16].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[17].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[17].turn_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20755 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[17].ttft_s", "statistic": "single", "unit": "s", "value": 4.177881577983499 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.70952229442078 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[17].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[17].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 160011 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[17].dropped_messages", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[17].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[18].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[18].turn_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20755 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[18].ttft_s", "statistic": "single", "unit": "s", "value": 4.161982175020967 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.62464709020834 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[18].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20001 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[18].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 180012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[18].dropped_messages", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[18].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[19].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[19].turn_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20757 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[19].ttft_s", "statistic": "single", "unit": "s", "value": 4.2007552070426755 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.99682203356258 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[19].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20002 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[19].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 200014 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[19].dropped_messages", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "engine_interval_cache_hit_rate", "scope": "machine_total", "source_field": "rolling_context_turns[19].engine_interval_cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 4.17597703699721 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 3.9647572839749046 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 3.9615960029768758 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 20 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "16/20" } ], "reasons": [ "16/20 requests passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 2, "configured_context_tokens": 262144, "context_tokens": 262144, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "47a03fd9bbfe612c42f992e302c0e2fe", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Finite lab cohort; configured 262144 context differs from measured inputs. Aggregate uses first-output to final-completion window. Whole-host memory sampled during speed only; missing counters remain unassessed. Broader coding quality and delivery acceptance unassessed." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "e4c63bec028344f58671b800", "measured_at": "2026-10-07T04:07:20.241809+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 85.19338989257812 }, { "basis": null, "n": 8, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 54.39598519797437 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": 8, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 63.45305588201154 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "preemptions", "scope": "machine_total", "source_field": "preemptions", "statistic": "single", "unit": "count", "value": 0.0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "correctness_probes_passed", "scope": "machine_total", "source_field": "correctness_probes_passed", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 126.01387458873165 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 48.078847492331505 }, { "basis": null, "n": null, "name": "correctness_probes_total", "scope": "machine_total", "source_field": "correctness_probes_total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": 8, "name": "decode_tps", "scope": "per_person", "source_field": "per_request_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 18.317248544411598 }, { "basis": null, "n": null, "name": "forced_output_tokens", "scope": "machine_total", "source_field": "forced_output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 93.47927856445312 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 63.45305588201154 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[0].total_s", "statistic": "single", "unit": "s", "value": 85.19329207600094 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 23.504804429920494 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40100 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 54.395742350956425 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[1].total_s", "statistic": "single", "unit": "s", "value": 84.81930265098345 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.796193310733134 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 63.3564414139837 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[2].total_s", "statistic": "single", "unit": "s", "value": 85.17781369196018 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 23.417409019492958 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[2].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40106 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 54.39515774499159 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[3].total_s", "statistic": "single", "unit": "s", "value": 84.817528176005 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.796850237517077 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40100 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 54.39622804499231 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[4].total_s", "statistic": "single", "unit": "s", "value": 84.8180215669563 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.797168767550378 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40100 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[5].ttft_s", "statistic": "single", "unit": "s", "value": 54.39632660499774 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[5].total_s", "statistic": "single", "unit": "s", "value": 84.81650915596401 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.798058300401898 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[5].user_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40098 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[5].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[5].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[5].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[6].ttft_s", "statistic": "single", "unit": "s", "value": 53.7162357950001 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[6].total_s", "statistic": "single", "unit": "s", "value": 84.77725951600587 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 16.451486100067715 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[6].user_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[6].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[6].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[6].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[7].ttft_s", "statistic": "single", "unit": "s", "value": 52.7525178049691 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[7].total_s", "statistic": "single", "unit": "s", "value": 84.73795960500138 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 15.976018189609135 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[7].user_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40106 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[7].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[7].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[7].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 8, "name": "decode_tps", "scope": "per_person", "source_field": "per_person_decode_median_tps", "statistic": "median", "unit": "tok/s", "value": 16.797009502533726 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "8/8" }, { "met": true, "name": "hidden-facts", "observed": "24/24" }, { "met": true, "name": "correctness-probes", "observed": "8/8" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 24, "skipped": null, "suite": "hidden-facts", "total": 24, "version": "1" }, { "passed": 8, "skipped": null, "suite": "correctness-probes", "total": 8, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "e2097959b1ca0142f2b93edc3ec411e474c0a6622800dd247af7a71bd9c39e83", "harness": { "name": "keep-team-load-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 23.504804429920494, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40100, "total_s": 85.19329207600094, "ttft_s": 63.45305588201154, "user_number": 0 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 16.796193310733134, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 84.81930265098345, "ttft_s": 54.395742350956425, "user_number": 1 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 23.417409019492958, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40106, "total_s": 85.17781369196018, "ttft_s": 63.3564414139837, "user_number": 2 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 16.796850237517077, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40100, "total_s": 84.817528176005, "ttft_s": 54.39515774499159, "user_number": 3 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 16.797168767550378, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40100, "total_s": 84.8180215669563, "ttft_s": 54.39622804499231, "user_number": 4 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 16.798058300401898, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40098, "total_s": 84.81650915596401, "ttft_s": 54.39632660499774, "user_number": 5 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 16.451486100067715, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 84.77725951600587, "ttft_s": 53.7162357950001, "user_number": 6 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 15.976018189609135, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40106, "total_s": 84.73795960500138, "ttft_s": 52.7525178049691, "user_number": 7 } ], "scenario": { "id": "team-load", "version": "1" }, "settings": { "concurrency": 8, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "plain", "thinking": false }, "setup_complete": false, "setup_id": "a4b96c03f67c006a12a64bbc7734ee71", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": { "max": 40106, "median": 40101.5, "min": 40098 }, "output_tokens": 512, "repetitions": null, "users": 8, "warmup": null } }, { "caveats": [ "Finite lab cohort; configured 262144 context differs from measured inputs. Aggregate uses first-output to final-completion window. Whole-host memory sampled during speed only; missing counters remain unassessed. Broader coding quality and delivery acceptance unassessed." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "e51d5cbb032741a59bd7cde5", "measured_at": "2026-10-07T05:50:54.427704+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 201.55434799194336 }, { "basis": null, "n": 20, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 113.83728480502032 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 60 }, { "basis": null, "n": 20, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 161.927070937003 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 39 }, { "basis": null, "n": null, "name": "preemptions", "scope": "machine_total", "source_field": "preemptions", "statistic": "single", "unit": "count", "value": 0.0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "correctness_probes_passed", "scope": "machine_total", "source_field": "correctness_probes_passed", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 71.6468224183778 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 50.80515554251064 }, { "basis": null, "n": null, "name": "correctness_probes_total", "scope": "machine_total", "source_field": "correctness_probes_total", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": 20, "name": "decode_tps", "scope": "per_person", "source_field": "per_request_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 6.892859712043292 }, { "basis": null, "n": null, "name": "forced_output_tokens", "scope": "machine_total", "source_field": "forced_output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 94.15425109863281 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 148.91227467701538 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[0].total_s", "statistic": "single", "unit": "s", "value": 201.11776147800265 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.788243177350017 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 93.12094456399791 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[1].total_s", "statistic": "single", "unit": "s", "value": 199.43900388700422 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.806333028027948 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[1].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40101 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 62.210124173027 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[2].total_s", "statistic": "single", "unit": "s", "value": 198.19672669901047 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.7577231176311012 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[2].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40099 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 93.12125973199727 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[3].total_s", "statistic": "single", "unit": "s", "value": 199.44018414401216 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.806293920165476 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[3].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 119.838628721016 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[4].total_s", "statistic": "single", "unit": "s", "value": 200.28996130701853 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.351666076553059 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[5].ttft_s", "statistic": "single", "unit": "s", "value": 126.1290391900111 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[5].total_s", "statistic": "single", "unit": "s", "value": 200.47818547097268 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[5].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.872977371777175 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[5].user_number", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[5].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[5].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[5].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[5].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[6].ttft_s", "statistic": "single", "unit": "s", "value": 60.640390462009236 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[6].total_s", "statistic": "single", "unit": "s", "value": 198.11452466598712 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[6].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.7170628712002025 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[6].user_number", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[6].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[6].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[6].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[6].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[7].ttft_s", "statistic": "single", "unit": "s", "value": 62.21007188403746 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[7].total_s", "statistic": "single", "unit": "s", "value": 198.19897111598402 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[7].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.757659653737058 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[7].user_number", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[7].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[7].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[7].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[7].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[8].ttft_s", "statistic": "single", "unit": "s", "value": 107.83594088902464 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[8].total_s", "statistic": "single", "unit": "s", "value": 199.93539294903167 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[8].decode_tps", "statistic": "single", "unit": "tok/s", "value": 5.548350056057446 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[8].user_number", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[8].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[8].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[8].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[8].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[9].ttft_s", "statistic": "single", "unit": "s", "value": 161.927070937003 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[9].total_s", "statistic": "single", "unit": "s", "value": 201.55384508304996 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[9].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.895321686208359 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[9].user_number", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[9].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40099 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[9].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[9].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[9].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[10].ttft_s", "statistic": "single", "unit": "s", "value": 137.90526522602886 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[10].total_s", "statistic": "single", "unit": "s", "value": 200.80344489502022 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[10].decode_tps", "statistic": "single", "unit": "tok/s", "value": 8.1242414754957 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[10].user_number", "statistic": "single", "unit": "count", "value": 10 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[10].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[10].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[10].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[10].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[11].ttft_s", "statistic": "single", "unit": "s", "value": 161.5899823899963 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[11].total_s", "statistic": "single", "unit": "s", "value": 201.52797863003798 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[11].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.794833194151922 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[11].user_number", "statistic": "single", "unit": "count", "value": 11 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[11].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40106 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[11].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[11].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[11].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[12].ttft_s", "statistic": "single", "unit": "s", "value": 58.90822813601699 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[12].total_s", "statistic": "single", "unit": "s", "value": 198.0308507910231 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[12].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.6730187387795943 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[12].user_number", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[12].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[12].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[12].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[12].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[13].ttft_s", "statistic": "single", "unit": "s", "value": 158.87563084997237 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[13].total_s", "statistic": "single", "unit": "s", "value": 201.4258340059896 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[13].decode_tps", "statistic": "single", "unit": "tok/s", "value": 12.009343366148817 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[13].user_number", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[13].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40104 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[13].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[13].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[13].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[14].ttft_s", "statistic": "single", "unit": "s", "value": 145.74805730499793 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[14].total_s", "statistic": "single", "unit": "s", "value": 201.01954388001468 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[14].decode_tps", "statistic": "single", "unit": "tok/s", "value": 9.245273316585209 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[14].user_number", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[14].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[14].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[14].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[14].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[15].ttft_s", "statistic": "single", "unit": "s", "value": 153.568589943985 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[15].total_s", "statistic": "single", "unit": "s", "value": 201.25991954101482 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[15].decode_tps", "statistic": "single", "unit": "tok/s", "value": 10.714735871650442 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[15].user_number", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[15].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[15].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[15].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[15].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[16].ttft_s", "statistic": "single", "unit": "s", "value": 91.52825919998577 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[16].total_s", "statistic": "single", "unit": "s", "value": 199.37849916698178 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[16].decode_tps", "statistic": "single", "unit": "tok/s", "value": 4.7380515811218835 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[16].user_number", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[16].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40099 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[16].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[16].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[16].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[17].ttft_s", "statistic": "single", "unit": "s", "value": 124.6268373390194 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[17].total_s", "statistic": "single", "unit": "s", "value": 200.43525092600612 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[17].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.740676605950218 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[17].user_number", "statistic": "single", "unit": "count", "value": 17 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[17].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40097 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[17].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[17].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[17].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[18].ttft_s", "statistic": "single", "unit": "s", "value": 62.209566475998145 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[18].total_s", "statistic": "single", "unit": "s", "value": 198.19883554801345 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[18].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.757649434304936 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[18].user_number", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[18].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40105 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[18].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[18].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[18].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[19].ttft_s", "statistic": "single", "unit": "s", "value": 62.21057442901656 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[19].total_s", "statistic": "single", "unit": "s", "value": 198.19657693902263 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[19].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.75773969796928 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[19].user_number", "statistic": "single", "unit": "count", "value": 19 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[19].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40102 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[19].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[19].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[19].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 20, "name": "decode_tps", "scope": "per_person", "source_field": "per_person_decode_median_tps", "statistic": "median", "unit": "tok/s", "value": 5.950008066305252 } ], "model": { "artifact_sha256": [ "648fe83171485f4be03ea58d86089e088c94c2a3dc88922233f9002f30d8bc8b", "8303cef7d375021d69dcafc60d3fad5f8186995b18b0db5e2a609aadeb8a4f51", "cb5c5978721a1e87406ae68f1cd5a855326e40ff8bc13832644c392115336987" ], "publisher_repo": "ornith-ai/Ornith-1.5-35B-A3B", "publisher_revision": "10fbf86fed7ecee4a061f8b499a618f46001cac1", "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "20/20" }, { "met": false, "name": "hidden-facts", "observed": "39/60" }, { "met": false, "name": "correctness-probes", "observed": "13/20" } ], "reasons": [ "39/60 Hidden facts found.", "13/20 probes passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 39, "skipped": null, "suite": "hidden-facts", "total": 60, "version": "1" }, { "passed": 13, "skipped": null, "suite": "correctness-probes", "total": 20, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "e2097959b1ca0142f2b93edc3ec411e474c0a6622800dd247af7a71bd9c39e83", "harness": { "name": "keep-team-load-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 9.788243177350017, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 201.11776147800265, "ttft_s": 148.91227467701538, "user_number": 0 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 4.806333028027948, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40101, "total_s": 199.43900388700422, "ttft_s": 93.12094456399791, "user_number": 1 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 3.7577231176311012, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40099, "total_s": 198.19672669901047, "ttft_s": 62.210124173027, "user_number": 2 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 4.806293920165476, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 199.44018414401216, "ttft_s": 93.12125973199727, "user_number": 3 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 6.351666076553059, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 200.28996130701853, "ttft_s": 119.838628721016, "user_number": 4 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 6.872977371777175, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 200.47818547097268, "ttft_s": 126.1290391900111, "user_number": 5 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.7170628712002025, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 198.11452466598712, "ttft_s": 60.640390462009236, "user_number": 6 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.757659653737058, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 198.19897111598402, "ttft_s": 62.21007188403746, "user_number": 7 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 5.548350056057446, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 199.93539294903167, "ttft_s": 107.83594088902464, "user_number": 8 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 12.895321686208359, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40099, "total_s": 201.55384508304996, "ttft_s": 161.927070937003, "user_number": 9 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 8.1242414754957, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 200.80344489502022, "ttft_s": 137.90526522602886, "user_number": 10 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 12.794833194151922, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40106, "total_s": 201.52797863003798, "ttft_s": 161.5899823899963, "user_number": 11 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.6730187387795943, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 198.0308507910231, "ttft_s": 58.90822813601699, "user_number": 12 }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 12.009343366148817, "effective_prefill_tps": null, "needle_correct": 0, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40104, "total_s": 201.4258340059896, "ttft_s": 158.87563084997237, "user_number": 13 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 9.245273316585209, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 201.01954388001468, "ttft_s": 145.74805730499793, "user_number": 14 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 10.714735871650442, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 201.25991954101482, "ttft_s": 153.568589943985, "user_number": 15 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 4.7380515811218835, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40099, "total_s": 199.37849916698178, "ttft_s": 91.52825919998577, "user_number": 16 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 6.740676605950218, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40097, "total_s": 200.43525092600612, "ttft_s": 124.6268373390194, "user_number": 17 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.757649434304936, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40105, "total_s": 198.19883554801345, "ttft_s": 62.209566475998145, "user_number": 18 }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.75773969796928, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40102, "total_s": 198.19657693902263, "ttft_s": 62.21057442901656, "user_number": 19 } ], "scenario": { "id": "team-load", "version": "1" }, "settings": { "concurrency": 20, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "plain", "thinking": false }, "setup_complete": false, "setup_id": "a167e4bafd3fd96d0b47cf9d6645473d", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": { "max": 40106, "median": 40102.5, "min": 40097 }, "output_tokens": 512, "repetitions": null, "users": 20, "warmup": null } }, { "caveats": [ "Startup, inference completion, concurrent decode and retrieval correctness are separate gates.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "e69ef75c67384e579ea52b69", "measured_at": "2026-10-07T03:53:44.084514+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/0" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "did_not_start" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "long-context-concurrency", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "a9b2513c7b4da2c5c126534490dcbefe", "standard": "pre-standard", "system": { "driver": null, "hardware": "unrecorded hardware", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "ec6866aca24743b390a07702", "measured_at": "2026-10-07T07:19:17.010613+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 679.786287624971 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 672.7909155070083 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 673.0887494969647 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 673.0887494969647 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 678.9544842210016 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 20.798758508471696 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 371.438387860214 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 123 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[0].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 673.0722804479883 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 670.6237579819863 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 679.0552248660242 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 15.181225492603613 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 372.80367273644305 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 129 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[1].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 670.6226384000038 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 672.7908825309714 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 679.2040119660087 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 19.64719428733437 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 371.6028360245963 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 127 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[2].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 672.7792680509738 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 672.7909155070083 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 679.3165866290219 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 20.381045491449914 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 371.60281781093056 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 134 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[3].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 672.7830149899819 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 672.7909744029748 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 679.7747627599747 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 21.47831410865346 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 371.6027852809057 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 151 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[4].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 672.7856175299967 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 72.01615260831932 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.9767775727278569 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 94.35124588012695 } ], "model": { "artifact_sha256": [ "648fe83171485f4be03ea58d86089e088c94c2a3dc88922233f9002f30d8bc8b", "8303cef7d375021d69dcafc60d3fad5f8186995b18b0db5e2a609aadeb8a4f51", "cb5c5978721a1e87406ae68f1cd5a855326e40ff8bc13832644c392115336987" ], "publisher_repo": "ornith-ai/Ornith-1.5-35B-A3B", "publisher_revision": "10fbf86fed7ecee4a061f8b499a618f46001cac1", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "5/5" }, { "met": true, "name": "hidden-facts", "observed": "15/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 673.0887494969647 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 15, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "77fda79ee1961c35a8f8a77e7d5a01d5f9ce414975e00e640a166e71e5f9093e", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 20.798758508471696, "effective_prefill_tps": 371.438387860214, "first_real_token_from_dispatch_s": 673.0722804479883, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 123, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 678.9544842210016, "ttft_s": 673.0887494969647, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 15.181225492603613, "effective_prefill_tps": 372.80367273644305, "first_real_token_from_dispatch_s": 670.6226384000038, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 129, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 679.0552248660242, "ttft_s": 670.6237579819863, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 19.64719428733437, "effective_prefill_tps": 371.6028360245963, "first_real_token_from_dispatch_s": 672.7792680509738, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 127, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 679.2040119660087, "ttft_s": 672.7908825309714, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 20.381045491449914, "effective_prefill_tps": 371.60281781093056, "first_real_token_from_dispatch_s": 672.7830149899819, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 134, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 679.3165866290219, "ttft_s": 672.7909155070083, "user_number": 4, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 21.47831410865346, "effective_prefill_tps": 371.6027852809057, "first_real_token_from_dispatch_s": 672.7856175299967, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 151, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 679.7747627599747, "ttft_s": 672.7909744029748, "user_number": 0, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "shared-record-groups-v2", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "shared", "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "bde5d43ff70430becda39144f69c34e0", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "shared", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Startup, inference completion, concurrent decode and retrieval correctness are separate gates.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "eca6f5cffdfa4f298b30cc18", "measured_at": "2026-10-07T15:53:35.008572+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/0" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "did_not_start" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "long-context-concurrency", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "a9b2513c7b4da2c5c126534490dcbefe", "standard": "pre-standard", "system": { "driver": null, "hardware": "unrecorded hardware", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Finite lab cohort; configured 262144 context differs from measured inputs. Aggregate uses first-output to final-completion window. Whole-host memory sampled during speed only; missing counters remain unassessed. Broader coding quality and delivery acceptance unassessed." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "ed509c2d71d243a1b4d18c7e", "measured_at": "2026-10-07T04:34:31.551279+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 16.037394046783447 }, { "basis": null, "n": 1, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 10.403273527976125 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 1, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 10.403273527976125 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "preemptions", "scope": "machine_total", "source_field": "preemptions", "statistic": "single", "unit": "count", "value": 0.0 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "mtp_acceptance_rate", "scope": "machine_total", "source_field": "mtp_acceptance_rate", "statistic": "single", "unit": "ratio", "value": 0.8339350180505415 }, { "basis": null, "n": null, "name": "correctness_probes_passed", "scope": "machine_total", "source_field": "correctness_probes_passed", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 90.69738626208539 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 31.925386288222413 }, { "basis": null, "n": null, "name": "correctness_probes_total", "scope": "machine_total", "source_field": "correctness_probes_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "single", "unit": "tok/s", "value": 90.69740914224184 }, { "basis": null, "n": 1, "name": "decode_tps", "scope": "per_person", "source_field": "per_request_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 90.69740914224184 }, { "basis": null, "n": null, "name": "forced_output_tokens", "scope": "machine_total", "source_field": "forced_output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 91.98944473266602 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "team_load_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 10.403273527976125 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "team_load_requests[0].total_s", "statistic": "single", "unit": "s", "value": 16.037392570986412 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "team_load_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 90.69740914224184 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "team_load_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "team_load_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40103 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "team_load_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 512 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "team_load_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "team_load_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 1, "name": "decode_tps", "scope": "per_person", "source_field": "per_person_decode_median_tps", "statistic": "median", "unit": "tok/s", "value": 90.69740914224184 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "1/1" }, { "met": true, "name": "hidden-facts", "observed": "3/3" }, { "met": true, "name": "correctness-probes", "observed": "1/1" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 3, "skipped": null, "suite": "hidden-facts", "total": 3, "version": "1" }, { "passed": 1, "skipped": null, "suite": "correctness-probes", "total": 1, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "e2097959b1ca0142f2b93edc3ec411e474c0a6622800dd247af7a71bd9c39e83", "harness": { "name": "keep-team-load-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 90.69740914224184, "effective_prefill_tps": null, "needle_correct": 3, "needle_total": 3, "output_tokens": 512, "prompt_tokens": 40103, "total_s": 16.037392570986412, "ttft_s": 10.403273527976125, "user_number": 0 } ], "scenario": { "id": "team-load", "version": "1" }, "settings": { "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "cf0bbc6439f3c0ed87c21751d34d5655", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": { "max": 40103, "median": 40103, "min": 40103 }, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Measured retrieval only, with no customer qualification threshold. Exact dense float32 array size excludes payloads and vector database overhead. One timing run; no reranking, ingestion, answer generation or GPU measurement.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "SentenceTransformers CPU", "recipe": "plain", "version": "6.1.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "fb23dc58fbe34ffca8dcd742", "measured_at": "2026-10-07T14:30:51.165607+00:00", "metrics": [ { "basis": null, "n": null, "name": "passages_per_second", "scope": "machine_total", "source_field": "passages_per_second", "statistic": "single", "unit": "passages/s", "value": 152.09623790026615 }, { "basis": null, "n": null, "name": "query_seconds", "scope": "per_person", "source_field": "query_seconds", "statistic": "single", "unit": "s", "value": 0.2951190830208361 }, { "basis": null, "n": null, "name": "recall_at_5", "scope": "machine_total", "source_field": "recall_at_5", "statistic": "single", "unit": "ratio", "value": 0.66 }, { "basis": null, "n": null, "name": "mrr", "scope": "machine_total", "source_field": "mrr", "statistic": "single", "unit": "ratio", "value": 0.4810901900976445 }, { "basis": null, "n": null, "name": "questions", "scope": "machine_total", "source_field": "questions", "statistic": "single", "unit": "count", "value": 100 }, { "basis": null, "n": null, "name": "passage_seconds", "scope": "machine_total", "source_field": "passage_seconds", "statistic": "single", "unit": "s", "value": 2.6299138329923153 }, { "basis": null, "n": null, "name": "vector_bytes", "scope": "machine_total", "source_field": "vector_bytes", "statistic": "single", "unit": "bytes", "value": 614400 }, { "basis": null, "n": null, "name": "passages", "scope": "machine_total", "source_field": "passages", "statistic": "single", "unit": "count", "value": 400 }, { "basis": null, "n": null, "name": "index_file_bytes", "scope": "machine_total", "source_field": "index_file_bytes", "statistic": "single", "unit": "bytes", "value": 614528 }, { "basis": null, "n": null, "name": "recall_at_10", "scope": "machine_total", "source_field": "recall_at_10", "statistic": "single", "unit": "ratio", "value": 0.74 }, { "basis": null, "n": null, "name": "truncated_passages", "scope": "machine_total", "source_field": "truncated_passages", "statistic": "single", "unit": "count", "value": 40 } ], "model": { "artifact_sha256": [], "publisher_repo": "sentence-transformers/all-MiniLM-L6-v2", "publisher_revision": "1110a243fdf4706b3f48f1d95db1a4f5529b4d41", "variant": "FP32 variant", "weights_dtype": "FP32", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-embedding-eval-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "embeddings", "version": "1" }, "settings": { "context_tokens": 256, "corpus_sha256": "d7fb630b25a7d5128c44491d1a8f21bde479711f79520f1ce95ea8bd8a0ae1f6", "dimensions": 384, "dtype": "float32" }, "setup_complete": false, "setup_id": "e53848969fe75d1dfb1dea5323eec4d8", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple Silicon laptop CPU, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "fcdbf0c4c05e4432a275612c", "measured_at": "2026-10-07T07:04:09.414932+00:00", "metrics": [ { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "wall_seconds", "statistic": "single", "unit": "s", "value": 682.3456255440251 }, { "basis": null, "n": 5, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 673.5430727319908 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": 5, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 676.9076247779885 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 650.2892515330459 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 681.654889094003 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 3.762078796283817 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 384.46122154195734 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 119 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[0].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 650.2877437130082 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[1].ttft_s", "statistic": "single", "unit": "s", "value": 665.1574827909935 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[1].total_s", "statistic": "single", "unit": "s", "value": 681.6506461239769 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 6.6087989186416065 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[1].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 375.8673794827603 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[1].user_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[1].output_tokens", "statistic": "single", "unit": "tokens", "value": 110 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[1].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[1].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[1].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 665.1531822580146 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[2].ttft_s", "statistic": "single", "unit": "s", "value": 673.5430727319908 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[2].total_s", "statistic": "single", "unit": "s", "value": 682.181508595997 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 13.544118616137988 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[2].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 371.18784250266015 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[2].output_tokens", "statistic": "single", "unit": "tokens", "value": 118 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[2].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[2].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[2].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 673.5334438770078 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[3].ttft_s", "statistic": "single", "unit": "s", "value": 676.9076247779885 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[3].total_s", "statistic": "single", "unit": "s", "value": 682.1814810289652 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 21.616061298388267 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[3].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 369.3428628197213 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[3].user_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[3].output_tokens", "statistic": "single", "unit": "tokens", "value": 115 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[3].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[3].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[3].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 676.8907029170077 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[4].ttft_s", "statistic": "single", "unit": "s", "value": 676.4575507600093 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[4].total_s", "statistic": "single", "unit": "s", "value": 682.3350459460053 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[4].decode_tps", "statistic": "single", "unit": "tok/s", "value": 20.75714162908789 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[4].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 369.58860126420234 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[4].user_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[4].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[4].output_tokens", "statistic": "single", "unit": "tokens", "value": 123 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[4].needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[4].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[4].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 676.4428287980263 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "aggregate_decode_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 18.099173248127094 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_end_to_end_tps", "statistic": "single", "unit": "tok/s", "value": 0.8573367778735114 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 94.3608627319336 } ], "model": { "artifact_sha256": [ "648fe83171485f4be03ea58d86089e088c94c2a3dc88922233f9002f30d8bc8b", "8303cef7d375021d69dcafc60d3fad5f8186995b18b0db5e2a609aadeb8a4f51", "cb5c5978721a1e87406ae68f1cd5a855326e40ff8bc13832644c392115336987" ], "publisher_repo": "ornith-ai/Ornith-1.5-35B-A3B", "publisher_revision": "10fbf86fed7ecee4a061f8b499a618f46001cac1", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "4/5" }, { "met": false, "name": "hidden-facts", "observed": "12/15" }, { "met": false, "name": "first-token-300-s-reference", "observed": 676.9076247779885 } ], "reasons": [ "One or more requests did not meet completion, exact retrieval, synthesis-presence, reasoning-off or token-accounting requirements." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 12, "skipped": null, "suite": "hidden-facts", "total": 15, "version": "1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0", "harness": { "name": "keep-long-context-concurrency-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 3.762078796283817, "effective_prefill_tps": 384.46122154195734, "first_real_token_from_dispatch_s": 650.2877437130082, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 119, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 681.654889094003, "ttft_s": 650.2892515330459, "user_number": 1, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 6.6087989186416065, "effective_prefill_tps": 375.8673794827603, "first_real_token_from_dispatch_s": 665.1531822580146, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 110, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 681.6506461239769, "ttft_s": 665.1574827909935, "user_number": 3, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 13.544118616137988, "effective_prefill_tps": 371.18784250266015, "first_real_token_from_dispatch_s": 673.5334438770078, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 118, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 682.181508595997, "ttft_s": 673.5430727319908, "user_number": 0, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 21.616061298388267, "effective_prefill_tps": 369.3428628197213, "first_real_token_from_dispatch_s": 676.8907029170077, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 115, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 682.1814810289652, "ttft_s": 676.9076247779885, "user_number": 2, "valid_json": true }, { "cached_prompt_tokens": null, "checks_passed": false, "completed": true, "decode_tps": 20.75714162908789, "effective_prefill_tps": 369.58860126420234, "first_real_token_from_dispatch_s": 676.4428287980263, "first_token_within_300_seconds": false, "needle_correct": 0, "needle_total": 3, "output_tokens": 123, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": false, "total_s": 682.3350459460053, "ttft_s": 676.4575507600093, "user_number": 4, "valid_json": false } ], "scenario": { "id": "long-context-concurrency", "version": "1" }, "settings": { "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "distinct-v1", "kv_dtype": "fp8", "max_output_tokens": 512, "scenario": "distinct", "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "bde5d43ff70430becda39144f69c34e0", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Bounded fictional tasks only. Unsupported cases are separately counted, never passes. Original publisher name with variant; exact variant repository and revision retained privately. Not customer acceptance.", "Publisher revision was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "ffd38b7a2ab2411ea1a39b6e", "measured_at": "2026-10-07T04:50:34.708410+00:00", "metrics": [ { "basis": null, "n": null, "name": "skipped_cases", "scope": "machine_total", "source_field": "skipped_cases", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "correctness_failures", "scope": "machine_total", "source_field": "correctness_failures", "statistic": "single", "unit": "count", "value": 18 }, { "basis": null, "n": null, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 0.06478124699788168 }, { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 130 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 103 }, { "basis": null, "n": null, "name": "api_cases_passed", "scope": "machine_total", "source_field": "api_cases_passed", "statistic": "single", "unit": "count", "value": 101 }, { "basis": null, "n": null, "name": "ttft_p95_s", "scope": "per_person", "source_field": "ttft_p95_s", "statistic": "p95", "unit": "s", "value": 0.11868229100946337 }, { "basis": null, "n": null, "name": "completion_p95_s", "scope": "per_person", "source_field": "completion_p95_s", "statistic": "p95", "unit": "s", "value": 0.6299726750003174 }, { "basis": null, "n": null, "name": "api_cases_total", "scope": "machine_total", "source_field": "api_cases_total", "statistic": "single", "unit": "count", "value": 124 }, { "basis": null, "n": null, "name": "completion_median_s", "scope": "per_person", "source_field": "completion_median_s", "statistic": "median", "unit": "s", "value": 0.17163818201515824 }, { "basis": null, "n": null, "name": "completeness_failures", "scope": "machine_total", "source_field": "completeness_failures", "statistic": "single", "unit": "count", "value": 9 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[0].passed", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[0].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[0].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[1].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[1].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[1].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[2].passed", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[2].total", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[2].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[3].passed", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[3].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[3].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[4].passed", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[4].total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[4].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[5].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[5].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[5].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[6].passed", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[6].total", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[6].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[7].passed", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[7].total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[7].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[8].passed", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[8].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[8].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[9].passed", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[9].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[9].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[10].passed", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[10].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[10].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[11].passed", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[11].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[11].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[12].passed", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[12].total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[12].skipped", "statistic": "single", "unit": "count", "value": 4 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "structured-tasks", "observed": "103/130" }, { "met": false, "name": "API-cases", "observed": "101/124" } ], "reasons": [ "103/130 tasks passed.", "101/124 API cases passed." ], "status": "partial" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 103, "skipped": null, "suite": "structured-tasks", "total": 130, "version": "core-v1" }, { "passed": 101, "skipped": 4, "suite": "API-cases", "total": 124, "version": "core-v1" } ], "record_date": "2026-10-07", "reproduce": { "driver": null, "fixture_sha256": "42c4dc34a066f791a62c9b1dbeb97943988c746309a43a464f0417a02ae9f1da", "harness": { "name": "keep-api-suite-run-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": "core-v1", "workload_version": "1" }, "requests": [], "scenario": { "id": "api-suite", "version": "1" }, "settings": { "concurrency": 1, "context_tokens": 262144, "kv_dtype": "fp8", "speculative_decoding": "with multi-token prediction" }, "setup_complete": false, "setup_id": "34e602e9ec2b44b5600f291faf6733df", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Lab cohort, not customer qualification. Reranking and application BM25 are separate recipe arms. Qdrant storage excludes model/document storage. Exact LLM and service pins are retained in the private source record.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "SentenceTransformers inside Open WebUI", "recipe": "plain", "version": "5.5.1" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "0934d05eaa71498ab5692df9", "measured_at": "2026-10-08T16:54:01.409699+00:00", "metrics": [ { "basis": null, "n": null, "name": "ingest_seconds", "scope": "machine_total", "source_field": "ingest_seconds", "statistic": "single", "unit": "s", "value": 85.652102137974 }, { "basis": null, "n": null, "name": "citation_correct", "scope": "machine_total", "source_field": "citation_correct", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "query_p50_seconds", "scope": "per_person", "source_field": "query_p50_seconds", "statistic": "median", "unit": "s", "value": 17.517732291016728 }, { "basis": null, "n": null, "name": "recall_at_5", "scope": "machine_total", "source_field": "recall_at_5", "statistic": "single", "unit": "ratio", "value": 0.3125 }, { "basis": null, "n": null, "name": "retrieval_p50_seconds", "scope": "per_person", "source_field": "retrieval_p50_seconds", "statistic": "median", "unit": "s", "value": 0.050306667981203645 }, { "basis": null, "n": null, "name": "retrieval_p95_seconds", "scope": "per_person", "source_field": "retrieval_p95_seconds", "statistic": "p95", "unit": "s", "value": 0.07371468999190256 }, { "basis": null, "n": null, "name": "qdrant_bytes", "scope": "machine_total", "source_field": "qdrant_bytes", "statistic": "single", "unit": "bytes", "value": 798007135 }, { "basis": null, "n": null, "name": "query_p95_seconds", "scope": "per_person", "source_field": "query_p95_seconds", "statistic": "p95", "unit": "s", "value": 73.60020717396401 }, { "basis": null, "n": null, "name": "mrr", "scope": "machine_total", "source_field": "mrr", "statistic": "single", "unit": "ratio", "value": 0.20467509920634921 }, { "basis": null, "n": null, "name": "answer_and_citation_correct", "scope": "machine_total", "source_field": "answer_and_citation_correct", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "completed_questions", "scope": "machine_total", "source_field": "completed_questions", "statistic": "single", "unit": "count", "value": 39 }, { "basis": null, "n": null, "name": "answer_correct", "scope": "machine_total", "source_field": "answer_correct", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "questions", "scope": "machine_total", "source_field": "questions", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "indexed_chunks", "scope": "machine_total", "source_field": "indexed_chunks", "statistic": "single", "unit": "count", "value": 833 }, { "basis": null, "n": null, "name": "pages", "scope": "machine_total", "source_field": "pages", "statistic": "single", "unit": "count", "value": 800 }, { "basis": null, "n": null, "name": "recall_at_10", "scope": "machine_total", "source_field": "recall_at_10", "statistic": "single", "unit": "ratio", "value": 0.625 }, { "basis": null, "n": null, "name": "retrieval_questions", "scope": "per_person", "source_field": "retrieval_questions", "statistic": "single", "unit": "count", "value": 32 } ], "model": { "artifact_sha256": [], "publisher_repo": "sentence-transformers/all-MiniLM-L6-v2", "publisher_revision": "1110a243fdf4706b3f48f1d95db1a4f5529b4d41", "variant": "float32 variant", "weights_dtype": "float32", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-08", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rag-trial-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rag", "version": "1" }, "settings": { "answer_model": "Qwen/Qwen3.6-35B-A3B", "answer_variant": "nvfp4; fp8 KV; with multi-token prediction", "corpus_sha256": "51e025e244e9860a3a9a91d69f910bc75ddcbc33c717d4ca8e957501a5a15ae2", "dimensions": 384, "dtype": "float32", "query_generation": true, "reranker_model": null, "thinking": false }, "setup_complete": false, "setup_id": "f221d0fcb7014a5f3147fd058edc5f17", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab cohort, not customer qualification. Reranking and application BM25 are separate recipe arms. Qdrant storage excludes model/document storage. Exact LLM and service pins are retained in the private source record.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Transformers text CPU", "recipe": "plain", "version": "5.19.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "166647151c374ea7852d42dc", "measured_at": "2026-10-08T18:46:36.441915+00:00", "metrics": [ { "basis": null, "n": null, "name": "ingest_seconds", "scope": "machine_total", "source_field": "ingest_seconds", "statistic": "single", "unit": "s", "value": 325.03059367701644 }, { "basis": null, "n": null, "name": "citation_correct", "scope": "machine_total", "source_field": "citation_correct", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "query_p50_seconds", "scope": "per_person", "source_field": "query_p50_seconds", "statistic": "median", "unit": "s", "value": 4.692133937031031 }, { "basis": null, "n": null, "name": "recall_at_5", "scope": "machine_total", "source_field": "recall_at_5", "statistic": "single", "unit": "ratio", "value": 0.6875 }, { "basis": null, "n": null, "name": "retrieval_p50_seconds", "scope": "per_person", "source_field": "retrieval_p50_seconds", "statistic": "median", "unit": "s", "value": 4.244149365986232 }, { "basis": null, "n": null, "name": "retrieval_p95_seconds", "scope": "per_person", "source_field": "retrieval_p95_seconds", "statistic": "p95", "unit": "s", "value": 6.479469697980676 }, { "basis": null, "n": null, "name": "qdrant_bytes", "scope": "machine_total", "source_field": "qdrant_bytes", "statistic": "single", "unit": "bytes", "value": 798011172 }, { "basis": null, "n": null, "name": "query_p95_seconds", "scope": "per_person", "source_field": "query_p95_seconds", "statistic": "p95", "unit": "s", "value": 7.578777200018521 }, { "basis": null, "n": null, "name": "mrr", "scope": "machine_total", "source_field": "mrr", "statistic": "single", "unit": "ratio", "value": 0.6948784722222222 }, { "basis": null, "n": null, "name": "answer_and_citation_correct", "scope": "machine_total", "source_field": "answer_and_citation_correct", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "completed_questions", "scope": "machine_total", "source_field": "completed_questions", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "answer_correct", "scope": "machine_total", "source_field": "answer_correct", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "questions", "scope": "machine_total", "source_field": "questions", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "indexed_chunks", "scope": "machine_total", "source_field": "indexed_chunks", "statistic": "single", "unit": "count", "value": 833 }, { "basis": null, "n": null, "name": "pages", "scope": "machine_total", "source_field": "pages", "statistic": "single", "unit": "count", "value": 800 }, { "basis": null, "n": null, "name": "recall_at_10", "scope": "machine_total", "source_field": "recall_at_10", "statistic": "single", "unit": "ratio", "value": 0.75 }, { "basis": null, "n": null, "name": "retrieval_questions", "scope": "per_person", "source_field": "retrieval_questions", "statistic": "single", "unit": "count", "value": 32 } ], "model": { "artifact_sha256": [], "publisher_repo": "google/embeddinggemma-2", "publisher_revision": "914f7f89142e33e77833254d9c9b90c3cef7303b", "variant": "float32 variant", "weights_dtype": "float32", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-08", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rag-trial-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rag", "version": "1" }, "settings": { "answer_model": "Qwen/Qwen3.6-35B-A3B", "answer_variant": "nvfp4; fp8 KV; with multi-token prediction", "corpus_sha256": "51e025e244e9860a3a9a91d69f910bc75ddcbc33c717d4ca8e957501a5a15ae2", "dimensions": 512, "dtype": "float32", "query_generation": false, "reranker_model": "BAAI/bge-reranker-v2-m3", "thinking": false }, "setup_complete": false, "setup_id": "404c7669f7c93ab3d26c64b15095c66d", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab cohort, not customer qualification. Reranking and application BM25 are separate recipe arms. Qdrant storage excludes model/document storage. Exact LLM and service pins are retained in the private source record.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Transformers text CPU", "recipe": "plain", "version": "5.19.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "1c681a2bddf141c39e04e36b", "measured_at": "2026-10-08T07:44:01.978917+00:00", "metrics": [ { "basis": null, "n": null, "name": "ingest_seconds", "scope": "machine_total", "source_field": "ingest_seconds", "statistic": "single", "unit": "s", "value": 321.1441417229944 }, { "basis": null, "n": null, "name": "citation_correct", "scope": "machine_total", "source_field": "citation_correct", "statistic": "single", "unit": "count", "value": 23 }, { "basis": null, "n": null, "name": "query_p50_seconds", "scope": "per_person", "source_field": "query_p50_seconds", "statistic": "median", "unit": "s", "value": 36.4604242190253 }, { "basis": null, "n": null, "name": "recall_at_5", "scope": "machine_total", "source_field": "recall_at_5", "statistic": "single", "unit": "ratio", "value": 0.65625 }, { "basis": null, "n": null, "name": "retrieval_p50_seconds", "scope": "per_person", "source_field": "retrieval_p50_seconds", "statistic": "median", "unit": "s", "value": 3.528093959961552 }, { "basis": null, "n": null, "name": "retrieval_p95_seconds", "scope": "per_person", "source_field": "retrieval_p95_seconds", "statistic": "p95", "unit": "s", "value": 5.414081858005375 }, { "basis": null, "n": null, "name": "qdrant_bytes", "scope": "machine_total", "source_field": "qdrant_bytes", "statistic": "single", "unit": "bytes", "value": 798011260 }, { "basis": null, "n": null, "name": "query_p95_seconds", "scope": "per_person", "source_field": "query_p95_seconds", "statistic": "p95", "unit": "s", "value": 98.6829914660193 }, { "basis": null, "n": null, "name": "mrr", "scope": "machine_total", "source_field": "mrr", "statistic": "single", "unit": "ratio", "value": 0.5622395833333333 }, { "basis": null, "n": null, "name": "answer_and_citation_correct", "scope": "machine_total", "source_field": "answer_and_citation_correct", "statistic": "single", "unit": "count", "value": 23 }, { "basis": null, "n": null, "name": "completed_questions", "scope": "machine_total", "source_field": "completed_questions", "statistic": "single", "unit": "count", "value": 39 }, { "basis": null, "n": null, "name": "answer_correct", "scope": "machine_total", "source_field": "answer_correct", "statistic": "single", "unit": "count", "value": 24 }, { "basis": null, "n": null, "name": "questions", "scope": "machine_total", "source_field": "questions", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "indexed_chunks", "scope": "machine_total", "source_field": "indexed_chunks", "statistic": "single", "unit": "count", "value": 833 }, { "basis": null, "n": null, "name": "pages", "scope": "machine_total", "source_field": "pages", "statistic": "single", "unit": "count", "value": 800 }, { "basis": null, "n": null, "name": "recall_at_10", "scope": "machine_total", "source_field": "recall_at_10", "statistic": "single", "unit": "ratio", "value": 0.71875 }, { "basis": null, "n": null, "name": "retrieval_questions", "scope": "per_person", "source_field": "retrieval_questions", "statistic": "single", "unit": "count", "value": 32 } ], "model": { "artifact_sha256": [], "publisher_repo": "google/embeddinggemma-2", "publisher_revision": "914f7f89142e33e77833254d9c9b90c3cef7303b", "variant": "float32 variant", "weights_dtype": "float32", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-08", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rag-trial-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rag", "version": "1" }, "settings": { "answer_model": "Qwen/Qwen3.6-35B-A3B", "answer_variant": "nvfp4; fp8 KV; with multi-token prediction", "corpus_sha256": "51e025e244e9860a3a9a91d69f910bc75ddcbc33c717d4ca8e957501a5a15ae2", "dimensions": 512, "dtype": "float32", "reranker_model": null }, "setup_complete": false, "setup_id": "89407b571387ffbb073bcb1e13e93271", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Synthetic media check only. CUDA allocator peaks are not total unified-memory residency.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NGC PyTorch / Diffusers video", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "1f246fdb5b864590a125538b", "measured_at": "2026-10-08T01:08:11.670694+00:00", "metrics": [ { "basis": null, "n": null, "name": "peak_allocated_gib", "scope": "machine_total", "source_field": "peak_allocated_gib", "statistic": "max", "unit": "GiB", "value": 24.05344533920288 }, { "basis": null, "n": null, "name": "request_seconds", "scope": "per_person", "source_field": "request_seconds", "statistic": "single", "unit": "s", "value": 28.912198656995315 }, { "basis": null, "n": null, "name": "fps", "scope": "machine_total", "source_field": "fps", "statistic": "single", "unit": "frames/s", "value": 24 }, { "basis": null, "n": null, "name": "frames", "scope": "machine_total", "source_field": "frames", "statistic": "single", "unit": "count", "value": 33 }, { "basis": null, "n": null, "name": "peak_reserved_gib", "scope": "machine_total", "source_field": "peak_reserved_gib", "statistic": "max", "unit": "GiB", "value": 24.396484375 }, { "basis": null, "n": null, "name": "clip_seconds", "scope": "machine_total", "source_field": "clip_seconds", "statistic": "single", "unit": "s", "value": 1.375 }, { "basis": null, "n": null, "name": "completed_clips", "scope": "machine_total", "source_field": "completed_clips", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "generation_seconds", "scope": "per_person", "source_field": "generation_seconds", "statistic": "single", "unit": "s", "value": 27.15624395798659 } ], "model": { "artifact_sha256": [], "publisher_repo": "Wan-AI/Wan2.2-TI2V-5B-Diffusers", "publisher_revision": "b8fff7315c768468a5333511427288870b2e9635", "variant": "F32 source / BF16 runtime / FP32 VAE variant", "weights_dtype": "F32 source / BF16 runtime / FP32 VAE", "weights_format": "unrecorded" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-08", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "media-video", "version": null }, "settings": { "dtype": "bf16", "fps": 24, "frames": 33, "height": 352, "speculative_decoding": "plain", "width": 640 }, "setup_complete": false, "setup_id": "e76730b16c4b8882d8aefee1b8e642ee", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab cohort, not customer qualification. Reranking and application BM25 are separate recipe arms. Qdrant storage excludes model/document storage. Exact LLM and service pins are retained in the private source record.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Transformers text CPU", "recipe": "plain", "version": "5.19.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "32df0b19901d4604b4832732", "measured_at": "2026-10-08T19:19:45.066538+00:00", "metrics": [ { "basis": null, "n": null, "name": "ingest_seconds", "scope": "machine_total", "source_field": "ingest_seconds", "statistic": "single", "unit": "s", "value": 325.03059367701644 }, { "basis": null, "n": null, "name": "citation_correct", "scope": "machine_total", "source_field": "citation_correct", "statistic": "single", "unit": "count", "value": 30 }, { "basis": null, "n": null, "name": "query_p50_seconds", "scope": "per_person", "source_field": "query_p50_seconds", "statistic": "median", "unit": "s", "value": 4.682209458027501 }, { "basis": null, "n": null, "name": "recall_at_5", "scope": "machine_total", "source_field": "recall_at_5", "statistic": "single", "unit": "ratio", "value": 0.6875 }, { "basis": null, "n": null, "name": "retrieval_p50_seconds", "scope": "per_person", "source_field": "retrieval_p50_seconds", "statistic": "median", "unit": "s", "value": 4.064931537024677 }, { "basis": null, "n": null, "name": "retrieval_p95_seconds", "scope": "per_person", "source_field": "retrieval_p95_seconds", "statistic": "p95", "unit": "s", "value": 6.370162591047119 }, { "basis": null, "n": null, "name": "qdrant_bytes", "scope": "machine_total", "source_field": "qdrant_bytes", "statistic": "single", "unit": "bytes", "value": 798011260 }, { "basis": null, "n": null, "name": "query_p95_seconds", "scope": "per_person", "source_field": "query_p95_seconds", "statistic": "p95", "unit": "s", "value": 7.410626160039101 }, { "basis": null, "n": null, "name": "mrr", "scope": "machine_total", "source_field": "mrr", "statistic": "single", "unit": "ratio", "value": 0.6948784722222222 }, { "basis": null, "n": null, "name": "answer_and_citation_correct", "scope": "machine_total", "source_field": "answer_and_citation_correct", "statistic": "single", "unit": "count", "value": 30 }, { "basis": null, "n": null, "name": "completed_questions", "scope": "machine_total", "source_field": "completed_questions", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "answer_correct", "scope": "machine_total", "source_field": "answer_correct", "statistic": "single", "unit": "count", "value": 30 }, { "basis": null, "n": null, "name": "questions", "scope": "machine_total", "source_field": "questions", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "indexed_chunks", "scope": "machine_total", "source_field": "indexed_chunks", "statistic": "single", "unit": "count", "value": 833 }, { "basis": null, "n": null, "name": "pages", "scope": "machine_total", "source_field": "pages", "statistic": "single", "unit": "count", "value": 800 }, { "basis": null, "n": null, "name": "recall_at_10", "scope": "machine_total", "source_field": "recall_at_10", "statistic": "single", "unit": "ratio", "value": 0.75 }, { "basis": null, "n": null, "name": "retrieval_questions", "scope": "per_person", "source_field": "retrieval_questions", "statistic": "single", "unit": "count", "value": 32 } ], "model": { "artifact_sha256": [], "publisher_repo": "google/embeddinggemma-2", "publisher_revision": "914f7f89142e33e77833254d9c9b90c3cef7303b", "variant": "float32 variant", "weights_dtype": "float32", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-08", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rag-trial-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rag", "version": "1" }, "settings": { "answer_format": "json-object-guided", "answer_model": "Qwen/Qwen3.6-35B-A3B", "answer_variant": "nvfp4; fp8 KV; with multi-token prediction", "corpus_sha256": "51e025e244e9860a3a9a91d69f910bc75ddcbc33c717d4ca8e957501a5a15ae2", "dimensions": 512, "dtype": "float32", "query_generation": false, "reranker_model": "BAAI/bge-reranker-v2-m3", "thinking": false }, "setup_complete": false, "setup_id": "19742f1dfd631bfd8c86bb138d5c4779", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab cohort, not customer qualification. Reranking and application BM25 are separate recipe arms. Qdrant storage excludes model/document storage. Exact LLM and service pins are retained in the private source record.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "SentenceTransformers inside Open WebUI", "recipe": "plain", "version": "5.5.1" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "45b9fcd99abd4ef2ac4998c5", "measured_at": "2026-10-08T07:06:16.538845+00:00", "metrics": [ { "basis": null, "n": null, "name": "ingest_seconds", "scope": "machine_total", "source_field": "ingest_seconds", "statistic": "single", "unit": "s", "value": 85.12173420999898 }, { "basis": null, "n": null, "name": "citation_correct", "scope": "machine_total", "source_field": "citation_correct", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "query_p50_seconds", "scope": "per_person", "source_field": "query_p50_seconds", "statistic": "median", "unit": "s", "value": 17.75983944098698 }, { "basis": null, "n": null, "name": "recall_at_5", "scope": "machine_total", "source_field": "recall_at_5", "statistic": "single", "unit": "ratio", "value": 0.3125 }, { "basis": null, "n": null, "name": "retrieval_p50_seconds", "scope": "per_person", "source_field": "retrieval_p50_seconds", "statistic": "median", "unit": "s", "value": 0.04783197201322764 }, { "basis": null, "n": null, "name": "retrieval_p95_seconds", "scope": "per_person", "source_field": "retrieval_p95_seconds", "statistic": "p95", "unit": "s", "value": 0.05823908204911277 }, { "basis": null, "n": null, "name": "qdrant_bytes", "scope": "machine_total", "source_field": "qdrant_bytes", "statistic": "single", "unit": "bytes", "value": 798007164 }, { "basis": null, "n": null, "name": "query_p95_seconds", "scope": "per_person", "source_field": "query_p95_seconds", "statistic": "p95", "unit": "s", "value": 40.64947353198659 }, { "basis": null, "n": null, "name": "mrr", "scope": "machine_total", "source_field": "mrr", "statistic": "single", "unit": "ratio", "value": 0.20467509920634921 }, { "basis": null, "n": null, "name": "answer_and_citation_correct", "scope": "machine_total", "source_field": "answer_and_citation_correct", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "completed_questions", "scope": "machine_total", "source_field": "completed_questions", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "answer_correct", "scope": "machine_total", "source_field": "answer_correct", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "questions", "scope": "machine_total", "source_field": "questions", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "indexed_chunks", "scope": "machine_total", "source_field": "indexed_chunks", "statistic": "single", "unit": "count", "value": 833 }, { "basis": null, "n": null, "name": "pages", "scope": "machine_total", "source_field": "pages", "statistic": "single", "unit": "count", "value": 800 }, { "basis": null, "n": null, "name": "recall_at_10", "scope": "machine_total", "source_field": "recall_at_10", "statistic": "single", "unit": "ratio", "value": 0.625 }, { "basis": null, "n": null, "name": "retrieval_questions", "scope": "per_person", "source_field": "retrieval_questions", "statistic": "single", "unit": "count", "value": 32 } ], "model": { "artifact_sha256": [], "publisher_repo": "sentence-transformers/all-MiniLM-L6-v2", "publisher_revision": "1110a243fdf4706b3f48f1d95db1a4f5529b4d41", "variant": "float32 variant", "weights_dtype": "float32", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-08", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rag-trial-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rag", "version": "1" }, "settings": { "answer_model": "Qwen/Qwen3.6-35B-A3B", "answer_variant": "nvfp4; fp8 KV; with multi-token prediction", "corpus_sha256": "51e025e244e9860a3a9a91d69f910bc75ddcbc33c717d4ca8e957501a5a15ae2", "dimensions": 384, "dtype": "float32", "reranker_model": null }, "setup_complete": false, "setup_id": "db25350cc4d07c249ed1f4b0a248bc41", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab cohort, not customer qualification. Reranking and application BM25 are separate recipe arms. Qdrant storage excludes model/document storage. Exact LLM and service pins are retained in the private source record.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Transformers text CPU", "recipe": "plain", "version": "5.19.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "641fc07e3f82447a87a1fd40", "measured_at": "2026-10-08T09:35:49.495129+00:00", "metrics": [ { "basis": null, "n": null, "name": "ingest_seconds", "scope": "machine_total", "source_field": "ingest_seconds", "statistic": "single", "unit": "s", "value": 326.194226238993 }, { "basis": null, "n": null, "name": "citation_correct", "scope": "machine_total", "source_field": "citation_correct", "statistic": "single", "unit": "count", "value": 25 }, { "basis": null, "n": null, "name": "query_p50_seconds", "scope": "per_person", "source_field": "query_p50_seconds", "statistic": "median", "unit": "s", "value": 25.61032143898774 }, { "basis": null, "n": null, "name": "recall_at_5", "scope": "machine_total", "source_field": "recall_at_5", "statistic": "single", "unit": "ratio", "value": 0.65625 }, { "basis": null, "n": null, "name": "retrieval_p50_seconds", "scope": "per_person", "source_field": "retrieval_p50_seconds", "statistic": "median", "unit": "s", "value": 3.2225852140109055 }, { "basis": null, "n": null, "name": "retrieval_p95_seconds", "scope": "per_person", "source_field": "retrieval_p95_seconds", "statistic": "p95", "unit": "s", "value": 6.230769980000332 }, { "basis": null, "n": null, "name": "qdrant_bytes", "scope": "machine_total", "source_field": "qdrant_bytes", "statistic": "single", "unit": "bytes", "value": 798002995 }, { "basis": null, "n": null, "name": "query_p95_seconds", "scope": "per_person", "source_field": "query_p95_seconds", "statistic": "p95", "unit": "s", "value": 78.08062157902168 }, { "basis": null, "n": null, "name": "mrr", "scope": "machine_total", "source_field": "mrr", "statistic": "single", "unit": "ratio", "value": 0.5951760912698413 }, { "basis": null, "n": null, "name": "answer_and_citation_correct", "scope": "machine_total", "source_field": "answer_and_citation_correct", "statistic": "single", "unit": "count", "value": 25 }, { "basis": null, "n": null, "name": "completed_questions", "scope": "machine_total", "source_field": "completed_questions", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "answer_correct", "scope": "machine_total", "source_field": "answer_correct", "statistic": "single", "unit": "count", "value": 25 }, { "basis": null, "n": null, "name": "questions", "scope": "machine_total", "source_field": "questions", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "indexed_chunks", "scope": "machine_total", "source_field": "indexed_chunks", "statistic": "single", "unit": "count", "value": 548 }, { "basis": null, "n": null, "name": "pages", "scope": "machine_total", "source_field": "pages", "statistic": "single", "unit": "count", "value": 800 }, { "basis": null, "n": null, "name": "recall_at_10", "scope": "machine_total", "source_field": "recall_at_10", "statistic": "single", "unit": "ratio", "value": 0.78125 }, { "basis": null, "n": null, "name": "retrieval_questions", "scope": "per_person", "source_field": "retrieval_questions", "statistic": "single", "unit": "count", "value": 32 } ], "model": { "artifact_sha256": [], "publisher_repo": "google/embeddinggemma-2", "publisher_revision": "914f7f89142e33e77833254d9c9b90c3cef7303b", "variant": "float32 variant", "weights_dtype": "float32", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-08", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rag-trial-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rag", "version": "1" }, "settings": { "answer_model": "Qwen/Qwen3.6-35B-A3B", "answer_variant": "nvfp4; fp8 KV; with multi-token prediction", "corpus_sha256": "51e025e244e9860a3a9a91d69f910bc75ddcbc33c717d4ca8e957501a5a15ae2", "dimensions": 512, "dtype": "float32", "reranker_model": null }, "setup_complete": false, "setup_id": "ab1d442d1a53ed9f8793ccb13936236f", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab cohort, not customer qualification. Reranking and application BM25 are separate recipe arms. Qdrant storage excludes model/document storage. Exact LLM and service pins are retained in the private source record.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Transformers text CPU", "recipe": "plain", "version": "5.19.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "740e0bebe94f4d1f87f8506d", "measured_at": "2026-10-08T17:22:27.944305+00:00", "metrics": [ { "basis": null, "n": null, "name": "ingest_seconds", "scope": "machine_total", "source_field": "ingest_seconds", "statistic": "single", "unit": "s", "value": 322.2389847160084 }, { "basis": null, "n": null, "name": "citation_correct", "scope": "machine_total", "source_field": "citation_correct", "statistic": "single", "unit": "count", "value": 27 }, { "basis": null, "n": null, "name": "query_p50_seconds", "scope": "per_person", "source_field": "query_p50_seconds", "statistic": "median", "unit": "s", "value": 31.31484205502784 }, { "basis": null, "n": null, "name": "recall_at_5", "scope": "machine_total", "source_field": "recall_at_5", "statistic": "single", "unit": "ratio", "value": 0.6875 }, { "basis": null, "n": null, "name": "retrieval_p50_seconds", "scope": "per_person", "source_field": "retrieval_p50_seconds", "statistic": "median", "unit": "s", "value": 5.233278140018228 }, { "basis": null, "n": null, "name": "retrieval_p95_seconds", "scope": "per_person", "source_field": "retrieval_p95_seconds", "statistic": "p95", "unit": "s", "value": 7.155100887990557 }, { "basis": null, "n": null, "name": "qdrant_bytes", "scope": "machine_total", "source_field": "qdrant_bytes", "statistic": "single", "unit": "bytes", "value": 798011230 }, { "basis": null, "n": null, "name": "query_p95_seconds", "scope": "per_person", "source_field": "query_p95_seconds", "statistic": "p95", "unit": "s", "value": 66.63510167103959 }, { "basis": null, "n": null, "name": "mrr", "scope": "machine_total", "source_field": "mrr", "statistic": "single", "unit": "ratio", "value": 0.6948784722222222 }, { "basis": null, "n": null, "name": "answer_and_citation_correct", "scope": "machine_total", "source_field": "answer_and_citation_correct", "statistic": "single", "unit": "count", "value": 27 }, { "basis": null, "n": null, "name": "completed_questions", "scope": "machine_total", "source_field": "completed_questions", "statistic": "single", "unit": "count", "value": 39 }, { "basis": null, "n": null, "name": "answer_correct", "scope": "machine_total", "source_field": "answer_correct", "statistic": "single", "unit": "count", "value": 27 }, { "basis": null, "n": null, "name": "questions", "scope": "machine_total", "source_field": "questions", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "indexed_chunks", "scope": "machine_total", "source_field": "indexed_chunks", "statistic": "single", "unit": "count", "value": 833 }, { "basis": null, "n": null, "name": "pages", "scope": "machine_total", "source_field": "pages", "statistic": "single", "unit": "count", "value": 800 }, { "basis": null, "n": null, "name": "recall_at_10", "scope": "machine_total", "source_field": "recall_at_10", "statistic": "single", "unit": "ratio", "value": 0.75 }, { "basis": null, "n": null, "name": "retrieval_questions", "scope": "per_person", "source_field": "retrieval_questions", "statistic": "single", "unit": "count", "value": 32 } ], "model": { "artifact_sha256": [], "publisher_repo": "google/embeddinggemma-2", "publisher_revision": "914f7f89142e33e77833254d9c9b90c3cef7303b", "variant": "float32 variant", "weights_dtype": "float32", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-08", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rag-trial-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rag", "version": "1" }, "settings": { "answer_model": "Qwen/Qwen3.6-35B-A3B", "answer_variant": "nvfp4; fp8 KV; with multi-token prediction", "corpus_sha256": "51e025e244e9860a3a9a91d69f910bc75ddcbc33c717d4ca8e957501a5a15ae2", "dimensions": 512, "dtype": "float32", "query_generation": true, "reranker_model": "BAAI/bge-reranker-v2-m3", "thinking": false }, "setup_complete": false, "setup_id": "715e087d6af2cecb2739e8230335ee61", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab cohort, not customer qualification. Reranking and application BM25 are separate recipe arms. Qdrant storage excludes model/document storage. Exact LLM and service pins are retained in the private source record.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Transformers text CPU", "recipe": "plain", "version": "5.19.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "7d583e3862b442e9a9cdcac1", "measured_at": "2026-10-08T09:05:24.757962+00:00", "metrics": [ { "basis": null, "n": null, "name": "ingest_seconds", "scope": "machine_total", "source_field": "ingest_seconds", "statistic": "single", "unit": "s", "value": 316.9917739779921 }, { "basis": null, "n": null, "name": "citation_correct", "scope": "machine_total", "source_field": "citation_correct", "statistic": "single", "unit": "count", "value": 26 }, { "basis": null, "n": null, "name": "query_p50_seconds", "scope": "per_person", "source_field": "query_p50_seconds", "statistic": "median", "unit": "s", "value": 22.218912621960044 }, { "basis": null, "n": null, "name": "recall_at_5", "scope": "machine_total", "source_field": "recall_at_5", "statistic": "single", "unit": "ratio", "value": 0.53125 }, { "basis": null, "n": null, "name": "retrieval_p50_seconds", "scope": "per_person", "source_field": "retrieval_p50_seconds", "statistic": "median", "unit": "s", "value": 1.6649779999861494 }, { "basis": null, "n": null, "name": "retrieval_p95_seconds", "scope": "per_person", "source_field": "retrieval_p95_seconds", "statistic": "p95", "unit": "s", "value": 2.0598465399816632 }, { "basis": null, "n": null, "name": "qdrant_bytes", "scope": "machine_total", "source_field": "qdrant_bytes", "statistic": "single", "unit": "bytes", "value": 798034112 }, { "basis": null, "n": null, "name": "query_p95_seconds", "scope": "per_person", "source_field": "query_p95_seconds", "statistic": "p95", "unit": "s", "value": 66.29657783202128 }, { "basis": null, "n": null, "name": "mrr", "scope": "machine_total", "source_field": "mrr", "statistic": "single", "unit": "ratio", "value": 0.47161458333333334 }, { "basis": null, "n": null, "name": "answer_and_citation_correct", "scope": "machine_total", "source_field": "answer_and_citation_correct", "statistic": "single", "unit": "count", "value": 26 }, { "basis": null, "n": null, "name": "completed_questions", "scope": "machine_total", "source_field": "completed_questions", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "answer_correct", "scope": "machine_total", "source_field": "answer_correct", "statistic": "single", "unit": "count", "value": 26 }, { "basis": null, "n": null, "name": "questions", "scope": "machine_total", "source_field": "questions", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "indexed_chunks", "scope": "machine_total", "source_field": "indexed_chunks", "statistic": "single", "unit": "count", "value": 1622 }, { "basis": null, "n": null, "name": "pages", "scope": "machine_total", "source_field": "pages", "statistic": "single", "unit": "count", "value": 800 }, { "basis": null, "n": null, "name": "recall_at_10", "scope": "machine_total", "source_field": "recall_at_10", "statistic": "single", "unit": "ratio", "value": 0.59375 }, { "basis": null, "n": null, "name": "retrieval_questions", "scope": "per_person", "source_field": "retrieval_questions", "statistic": "single", "unit": "count", "value": 32 } ], "model": { "artifact_sha256": [], "publisher_repo": "google/embeddinggemma-2", "publisher_revision": "914f7f89142e33e77833254d9c9b90c3cef7303b", "variant": "float32 variant", "weights_dtype": "float32", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-08", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rag-trial-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rag", "version": "1" }, "settings": { "answer_model": "Qwen/Qwen3.6-35B-A3B", "answer_variant": "nvfp4; fp8 KV; with multi-token prediction", "corpus_sha256": "51e025e244e9860a3a9a91d69f910bc75ddcbc33c717d4ca8e957501a5a15ae2", "dimensions": 512, "dtype": "float32", "reranker_model": null }, "setup_complete": false, "setup_id": "a70f9c48272b09d7ee5d616b10ae2dde", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab cohort, not customer qualification. Reranking and application BM25 are separate recipe arms. Qdrant storage excludes model/document storage. Exact LLM and service pins are retained in the private source record.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Transformers text CPU", "recipe": "plain", "version": "5.19.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "8938e1a862644c4ca5e50a02", "measured_at": "2026-10-08T07:23:04.248664+00:00", "metrics": [ { "basis": null, "n": null, "name": "ingest_seconds", "scope": "machine_total", "source_field": "ingest_seconds", "statistic": "single", "unit": "s", "value": 82.63955245102989 }, { "basis": null, "n": null, "name": "citation_correct", "scope": "machine_total", "source_field": "citation_correct", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "query_p50_seconds", "scope": "per_person", "source_field": "query_p50_seconds", "statistic": "median", "unit": "s", "value": 20.672890101966914 }, { "basis": null, "n": null, "name": "recall_at_5", "scope": "machine_total", "source_field": "recall_at_5", "statistic": "single", "unit": "ratio", "value": 0.3125 }, { "basis": null, "n": null, "name": "retrieval_p50_seconds", "scope": "per_person", "source_field": "retrieval_p50_seconds", "statistic": "median", "unit": "s", "value": 0.4682569800061174 }, { "basis": null, "n": null, "name": "retrieval_p95_seconds", "scope": "per_person", "source_field": "retrieval_p95_seconds", "statistic": "p95", "unit": "s", "value": 0.7427361750160344 }, { "basis": null, "n": null, "name": "qdrant_bytes", "scope": "machine_total", "source_field": "qdrant_bytes", "statistic": "single", "unit": "bytes", "value": 798007105 }, { "basis": null, "n": null, "name": "query_p95_seconds", "scope": "per_person", "source_field": "query_p95_seconds", "statistic": "p95", "unit": "s", "value": 65.26910257001873 }, { "basis": null, "n": null, "name": "mrr", "scope": "machine_total", "source_field": "mrr", "statistic": "single", "unit": "ratio", "value": 0.20467509920634921 }, { "basis": null, "n": null, "name": "answer_and_citation_correct", "scope": "machine_total", "source_field": "answer_and_citation_correct", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "completed_questions", "scope": "machine_total", "source_field": "completed_questions", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "answer_correct", "scope": "machine_total", "source_field": "answer_correct", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "questions", "scope": "machine_total", "source_field": "questions", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "indexed_chunks", "scope": "machine_total", "source_field": "indexed_chunks", "statistic": "single", "unit": "count", "value": 833 }, { "basis": null, "n": null, "name": "pages", "scope": "machine_total", "source_field": "pages", "statistic": "single", "unit": "count", "value": 800 }, { "basis": null, "n": null, "name": "recall_at_10", "scope": "machine_total", "source_field": "recall_at_10", "statistic": "single", "unit": "ratio", "value": 0.625 }, { "basis": null, "n": null, "name": "retrieval_questions", "scope": "per_person", "source_field": "retrieval_questions", "statistic": "single", "unit": "count", "value": 32 } ], "model": { "artifact_sha256": [], "publisher_repo": "sentence-transformers/all-MiniLM-L6-v2", "publisher_revision": "1110a243fdf4706b3f48f1d95db1a4f5529b4d41", "variant": "float32 variant", "weights_dtype": "float32", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-08", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rag-trial-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rag", "version": "1" }, "settings": { "answer_model": "Qwen/Qwen3.6-35B-A3B", "answer_variant": "nvfp4; fp8 KV; with multi-token prediction", "corpus_sha256": "51e025e244e9860a3a9a91d69f910bc75ddcbc33c717d4ca8e957501a5a15ae2", "dimensions": 384, "dtype": "float32", "reranker_model": null }, "setup_complete": false, "setup_id": "384b5f2da363bf2e1251c6a67db2c9e2", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Native file/binary/library provenance checked separately; no OCI identity invented. Synthetic task pass is not broader intelligence or customer acceptance.", "Publisher revision was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Mirai S CUDA", "recipe": "with multi-token prediction", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "8c806d70a9584a5ea4c955cb", "measured_at": "2026-10-08T03:06:37.238561+00:00", "metrics": [ { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 16 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": null, "variant": "Mirai S trellis with Q8_0 MTP GGUF variant", "weights_dtype": "Mirai S trellis with Q8_0 MTP", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "16/16" }, { "met": true, "name": "structured-tasks", "observed": "16/16" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 16, "skipped": null, "suite": "structured-tasks", "total": 16, "version": null } ], "record_date": "2026-10-08", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": { "concurrency": 1, "context_tokens": 262144, "kv_dtype": "Q8_0", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction" }, "setup_complete": false, "setup_id": "d00f51ba1522542db97dd753e6b4453e", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA RTX 3060, nominal 12 GB, one GPU", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Synthetic media check only. CUDA allocator peaks are not total unified-memory residency.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NGC PyTorch / Diffusers video", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "93a2b3aa256047e589249f64", "measured_at": "2026-10-08T00:38:20.026137+00:00", "metrics": [ { "basis": null, "n": null, "name": "peak_allocated_gib", "scope": "machine_total", "source_field": "peak_allocated_gib", "statistic": "max", "unit": "GiB", "value": 28.40930461883545 }, { "basis": null, "n": null, "name": "request_seconds", "scope": "per_person", "source_field": "request_seconds", "statistic": "single", "unit": "s", "value": 1236.160940679023 }, { "basis": null, "n": null, "name": "fps", "scope": "machine_total", "source_field": "fps", "statistic": "single", "unit": "frames/s", "value": 24 }, { "basis": null, "n": null, "name": "frames", "scope": "machine_total", "source_field": "frames", "statistic": "single", "unit": "count", "value": 121 }, { "basis": null, "n": null, "name": "peak_reserved_gib", "scope": "machine_total", "source_field": "peak_reserved_gib", "statistic": "max", "unit": "GiB", "value": 30.404296875 }, { "basis": null, "n": null, "name": "clip_seconds", "scope": "machine_total", "source_field": "clip_seconds", "statistic": "single", "unit": "s", "value": 5.042 }, { "basis": null, "n": null, "name": "completed_clips", "scope": "machine_total", "source_field": "completed_clips", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "generation_seconds", "scope": "per_person", "source_field": "generation_seconds", "statistic": "single", "unit": "s", "value": 1225.5994232270168 } ], "model": { "artifact_sha256": [], "publisher_repo": "Wan-AI/Wan2.2-TI2V-5B-Diffusers", "publisher_revision": "b8fff7315c768468a5333511427288870b2e9635", "variant": "F32 source / BF16 runtime / FP32 VAE variant", "weights_dtype": "F32 source / BF16 runtime / FP32 VAE", "weights_format": "unrecorded" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-08", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "media-video", "version": null }, "settings": { "dtype": "bf16", "fps": 24, "frames": 121, "height": 704, "speculative_decoding": "plain", "width": 1280 }, "setup_complete": false, "setup_id": "593bee76ce616307f346b1de571a8f10", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab cohort, not customer qualification. Reranking and application BM25 are separate recipe arms. Qdrant storage excludes model/document storage. Exact LLM and service pins are retained in the private source record.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Transformers text CPU", "recipe": "plain", "version": "5.19.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "bb29199707ca422094795adb", "measured_at": "2026-10-08T10:08:18.102251+00:00", "metrics": [ { "basis": null, "n": null, "name": "ingest_seconds", "scope": "machine_total", "source_field": "ingest_seconds", "statistic": "single", "unit": "s", "value": 322.2998780099442 }, { "basis": null, "n": null, "name": "citation_correct", "scope": "machine_total", "source_field": "citation_correct", "statistic": "single", "unit": "count", "value": 23 }, { "basis": null, "n": null, "name": "query_p50_seconds", "scope": "per_person", "source_field": "query_p50_seconds", "statistic": "median", "unit": "s", "value": 35.694118414015975 }, { "basis": null, "n": null, "name": "recall_at_5", "scope": "machine_total", "source_field": "recall_at_5", "statistic": "single", "unit": "ratio", "value": 0.65625 }, { "basis": null, "n": null, "name": "retrieval_p50_seconds", "scope": "per_person", "source_field": "retrieval_p50_seconds", "statistic": "median", "unit": "s", "value": 4.310385440010577 }, { "basis": null, "n": null, "name": "retrieval_p95_seconds", "scope": "per_person", "source_field": "retrieval_p95_seconds", "statistic": "p95", "unit": "s", "value": 5.986723428010009 }, { "basis": null, "n": null, "name": "qdrant_bytes", "scope": "machine_total", "source_field": "qdrant_bytes", "statistic": "single", "unit": "bytes", "value": 798011231 }, { "basis": null, "n": null, "name": "query_p95_seconds", "scope": "per_person", "source_field": "query_p95_seconds", "statistic": "p95", "unit": "s", "value": 97.27227598702302 }, { "basis": null, "n": null, "name": "mrr", "scope": "machine_total", "source_field": "mrr", "statistic": "single", "unit": "ratio", "value": 0.5622395833333333 }, { "basis": null, "n": null, "name": "answer_and_citation_correct", "scope": "machine_total", "source_field": "answer_and_citation_correct", "statistic": "single", "unit": "count", "value": 23 }, { "basis": null, "n": null, "name": "completed_questions", "scope": "machine_total", "source_field": "completed_questions", "statistic": "single", "unit": "count", "value": 39 }, { "basis": null, "n": null, "name": "answer_correct", "scope": "machine_total", "source_field": "answer_correct", "statistic": "single", "unit": "count", "value": 23 }, { "basis": null, "n": null, "name": "questions", "scope": "machine_total", "source_field": "questions", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "indexed_chunks", "scope": "machine_total", "source_field": "indexed_chunks", "statistic": "single", "unit": "count", "value": 833 }, { "basis": null, "n": null, "name": "pages", "scope": "machine_total", "source_field": "pages", "statistic": "single", "unit": "count", "value": 800 }, { "basis": null, "n": null, "name": "recall_at_10", "scope": "machine_total", "source_field": "recall_at_10", "statistic": "single", "unit": "ratio", "value": 0.71875 }, { "basis": null, "n": null, "name": "retrieval_questions", "scope": "per_person", "source_field": "retrieval_questions", "statistic": "single", "unit": "count", "value": 32 } ], "model": { "artifact_sha256": [], "publisher_repo": "google/embeddinggemma-2", "publisher_revision": "914f7f89142e33e77833254d9c9b90c3cef7303b", "variant": "float32 variant", "weights_dtype": "float32", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-08", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rag-trial-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rag", "version": "1" }, "settings": { "answer_model": "Qwen/Qwen3.6-35B-A3B", "answer_variant": "nvfp4; fp8 KV; with multi-token prediction", "corpus_sha256": "51e025e244e9860a3a9a91d69f910bc75ddcbc33c717d4ca8e957501a5a15ae2", "dimensions": 512, "dtype": "float32", "reranker_model": null }, "setup_complete": false, "setup_id": "ac1e5ff4dd660f40f8bbae87f1b7a7ff", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Measured short throughput only; output hits the ceiling and is not answer completeness. Different quantization/runtime from compared unified-memory baselines.", "Publisher revision was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Mirai S CUDA", "recipe": "with multi-token prediction", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "c499ad73eb964222820e5320", "measured_at": "2026-10-08T04:28:10.783648+00:00", "metrics": [ { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 37.34571069231172 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_ttft_mean_s", "scope": "per_person", "source_field": "code_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.3015949793237572 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 41.47951138364672 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.3130799546682586 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 13.387645721435547 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": null, "variant": "Mirai S trellis with Q8_0 MTP GGUF variant", "weights_dtype": "Mirai S trellis with Q8_0 MTP", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-08", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 262144, "kv_dtype": "Q8_0", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction" }, "setup_complete": false, "setup_id": "d00f51ba1522542db97dd753e6b4453e", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA RTX 3060, nominal 12 GB, one GPU", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Correct retrieval and completion; first token exceeds the 300-second reporting bar. Configured window is 262144, not a filled 262144-token input; synthesis semantics and broader quality unassessed.", "Publisher revision was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Mirai S CUDA", "recipe": "with multi-token prediction", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "caf8d447ceff44c9bd32727f", "measured_at": "2026-10-08T04:24:25.168145+00:00", "metrics": [ { "basis": null, "n": 1, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 1282.1138055379852 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 1, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 1282.1138055379852 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 1282.1138055379852 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 1325.3815719729755 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.565420153286098 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 194.9990702230161 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[0].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 1282.1138055379852 }, { "basis": null, "n": null, "name": "cached_prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].cached_prompt_tokens", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.565420153286098 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 12.874828338623047 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": null, "variant": "Mirai S trellis with Q8_0 MTP GGUF variant", "weights_dtype": "Mirai S trellis with Q8_0 MTP", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "1/1" }, { "met": true, "name": "hidden-facts", "observed": "3/3" }, { "met": false, "name": "first-token-300-s-reference", "observed": 1282.1138055379852 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 3, "skipped": null, "suite": "hidden-facts", "total": 3, "version": null } ], "record_date": "2026-10-08", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": 0, "checks_passed": true, "completed": true, "decode_tps": 2.565420153286098, "effective_prefill_tps": 194.9990702230161, "first_real_token_from_dispatch_s": 1282.1138055379852, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 1325.3815719729755, "ttft_s": 1282.1138055379852, "user_number": 0, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": null }, "settings": { "cache_state": "cold", "concurrency": 1, "context_tokens": 250000, "fixture_variant": "original-tokenizer-needles-v1", "kv_dtype": "Q8_0", "max_output_tokens": 512, "scenario": "distinct", "speculative_decoding": "with multi-token prediction" }, "setup_complete": false, "setup_id": "52fd8db0d13a1f9802a39424ca764837", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA RTX 3060, nominal 12 GB, one GPU", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Lab cohort, not customer qualification. Reranking and application BM25 are separate recipe arms. Qdrant storage excludes model/document storage. Exact LLM and service pins are retained in the private source record.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Transformers text CPU", "recipe": "plain", "version": "5.19.0" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "e2b8dc39801f4b2caac56444", "measured_at": "2026-10-08T08:32:36.078864+00:00", "metrics": [ { "basis": null, "n": null, "name": "ingest_seconds", "scope": "machine_total", "source_field": "ingest_seconds", "statistic": "single", "unit": "s", "value": 323.04162843100494 }, { "basis": null, "n": null, "name": "citation_correct", "scope": "machine_total", "source_field": "citation_correct", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "query_p50_seconds", "scope": "per_person", "source_field": "query_p50_seconds", "statistic": "median", "unit": "s", "value": 26.892312394978944 }, { "basis": null, "n": null, "name": "recall_at_5", "scope": "machine_total", "source_field": "recall_at_5", "statistic": "single", "unit": "ratio", "value": 0.6875 }, { "basis": null, "n": null, "name": "retrieval_p50_seconds", "scope": "per_person", "source_field": "retrieval_p50_seconds", "statistic": "median", "unit": "s", "value": 4.009056763025001 }, { "basis": null, "n": null, "name": "retrieval_p95_seconds", "scope": "per_person", "source_field": "retrieval_p95_seconds", "statistic": "p95", "unit": "s", "value": 6.35553997696843 }, { "basis": null, "n": null, "name": "qdrant_bytes", "scope": "machine_total", "source_field": "qdrant_bytes", "statistic": "single", "unit": "bytes", "value": 798011201 }, { "basis": null, "n": null, "name": "query_p95_seconds", "scope": "per_person", "source_field": "query_p95_seconds", "statistic": "p95", "unit": "s", "value": 74.42034314799821 }, { "basis": null, "n": null, "name": "mrr", "scope": "machine_total", "source_field": "mrr", "statistic": "single", "unit": "ratio", "value": 0.6948784722222222 }, { "basis": null, "n": null, "name": "answer_and_citation_correct", "scope": "machine_total", "source_field": "answer_and_citation_correct", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "completed_questions", "scope": "machine_total", "source_field": "completed_questions", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "answer_correct", "scope": "machine_total", "source_field": "answer_correct", "statistic": "single", "unit": "count", "value": 29 }, { "basis": null, "n": null, "name": "questions", "scope": "machine_total", "source_field": "questions", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "indexed_chunks", "scope": "machine_total", "source_field": "indexed_chunks", "statistic": "single", "unit": "count", "value": 833 }, { "basis": null, "n": null, "name": "pages", "scope": "machine_total", "source_field": "pages", "statistic": "single", "unit": "count", "value": 800 }, { "basis": null, "n": null, "name": "recall_at_10", "scope": "machine_total", "source_field": "recall_at_10", "statistic": "single", "unit": "ratio", "value": 0.75 }, { "basis": null, "n": null, "name": "retrieval_questions", "scope": "per_person", "source_field": "retrieval_questions", "statistic": "single", "unit": "count", "value": 32 } ], "model": { "artifact_sha256": [], "publisher_repo": "google/embeddinggemma-2", "publisher_revision": "914f7f89142e33e77833254d9c9b90c3cef7303b", "variant": "float32 variant", "weights_dtype": "float32", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-08", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rag-trial-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rag", "version": "1" }, "settings": { "answer_model": "Qwen/Qwen3.6-35B-A3B", "answer_variant": "nvfp4; fp8 KV; with multi-token prediction", "corpus_sha256": "51e025e244e9860a3a9a91d69f910bc75ddcbc33c717d4ca8e957501a5a15ae2", "dimensions": 512, "dtype": "float32", "reranker_model": "BAAI/bge-reranker-v2-m3" }, "setup_complete": false, "setup_id": "2583532c8593653187e9aba28273b681", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Synthetic media check only. CUDA allocator peaks are not total unified-memory residency.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NGC PyTorch / Diffusers video", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "f19c0f4be02b4891a39dd1e1", "measured_at": "2026-10-08T00:28:21.136163+00:00", "metrics": [ { "basis": null, "n": null, "name": "peak_allocated_gib", "scope": "machine_total", "source_field": "peak_allocated_gib", "statistic": "max", "unit": "GiB", "value": 15.065889358520508 }, { "basis": null, "n": null, "name": "request_seconds", "scope": "per_person", "source_field": "request_seconds", "statistic": "single", "unit": "s", "value": 20.752855163998902 }, { "basis": null, "n": null, "name": "fps", "scope": "machine_total", "source_field": "fps", "statistic": "single", "unit": "frames/s", "value": 15 }, { "basis": null, "n": null, "name": "frames", "scope": "machine_total", "source_field": "frames", "statistic": "single", "unit": "count", "value": 33 }, { "basis": null, "n": null, "name": "peak_reserved_gib", "scope": "machine_total", "source_field": "peak_reserved_gib", "statistic": "max", "unit": "GiB", "value": 15.77734375 }, { "basis": null, "n": null, "name": "clip_seconds", "scope": "machine_total", "source_field": "clip_seconds", "statistic": "single", "unit": "s", "value": 2.2 }, { "basis": null, "n": null, "name": "completed_clips", "scope": "machine_total", "source_field": "completed_clips", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "generation_seconds", "scope": "per_person", "source_field": "generation_seconds", "statistic": "single", "unit": "s", "value": 18.61537449102616 } ], "model": { "artifact_sha256": [], "publisher_repo": "Wan-AI/Wan2.1-T2V-1.3B-Diffusers", "publisher_revision": "0fad780a534b6463e45facd96134c9f345acfa5b", "variant": "F32 source / BF16 runtime / FP32 VAE variant", "weights_dtype": "F32 source / BF16 runtime / FP32 VAE", "weights_format": "unrecorded" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-08", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "media-video", "version": null }, "settings": { "dtype": "bf16", "fps": 15, "frames": 33, "height": 240, "speculative_decoding": "plain", "width": 416 }, "setup_complete": false, "setup_id": "d54a85e6972eccd5996f718537658f5e", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Synthetic media check only. CUDA allocator peaks are not total unified-memory residency.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NGC PyTorch / Diffusers video", "recipe": "plain", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "f9293514116a49a3b9adef95", "measured_at": "2026-10-08T00:11:29.930701+00:00", "metrics": [ { "basis": null, "n": null, "name": "peak_allocated_gib", "scope": "machine_total", "source_field": "peak_allocated_gib", "statistic": "max", "unit": "GiB", "value": 15.673772811889648 }, { "basis": null, "n": null, "name": "request_seconds", "scope": "per_person", "source_field": "request_seconds", "statistic": "single", "unit": "s", "value": 567.7240323090227 }, { "basis": null, "n": null, "name": "fps", "scope": "machine_total", "source_field": "fps", "statistic": "single", "unit": "frames/s", "value": 15 }, { "basis": null, "n": null, "name": "frames", "scope": "machine_total", "source_field": "frames", "statistic": "single", "unit": "count", "value": 81 }, { "basis": null, "n": null, "name": "peak_reserved_gib", "scope": "machine_total", "source_field": "peak_reserved_gib", "statistic": "max", "unit": "GiB", "value": 16.748046875 }, { "basis": null, "n": null, "name": "clip_seconds", "scope": "machine_total", "source_field": "clip_seconds", "statistic": "single", "unit": "s", "value": 5.4 }, { "basis": null, "n": null, "name": "completed_clips", "scope": "machine_total", "source_field": "completed_clips", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "generation_seconds", "scope": "per_person", "source_field": "generation_seconds", "statistic": "single", "unit": "s", "value": 562.5696565919789 } ], "model": { "artifact_sha256": [], "publisher_repo": "Wan-AI/Wan2.1-T2V-1.3B-Diffusers", "publisher_revision": "0fad780a534b6463e45facd96134c9f345acfa5b", "variant": "F32 source / BF16 runtime / FP32 VAE variant", "weights_dtype": "F32 source / BF16 runtime / FP32 VAE", "weights_format": "unrecorded" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-08", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "media-video", "version": null }, "settings": { "dtype": "bf16", "fps": 15, "frames": 81, "height": 480, "speculative_decoding": "plain", "width": 832 }, "setup_complete": false, "setup_id": "398a7ad77ea2e0e12fddd97e9b590bb7", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Speed only; outputs reach the token ceiling and do not establish correctness. Whole-Mac memory includes existing applications. Filled contexts observed whole-Mac swap growth; a memory-stable recipe, managed lifecycle and delivery remain unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "136c3c4112ce4092b34f7835", "measured_at": "2026-10-09T20:03:41.006639+00:00", "metrics": [ { "basis": null, "n": null, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 0.13782647900370648 }, { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 28.364087811849625 }, { "basis": null, "n": null, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 0.29438474999915343 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 56.456619224524324 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.16283139566803584 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 13.367050170898438 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 2, "context_tokens": 4096, "kv_dtype": "q8_0", "max_output_tokens": 512, "thinking": false }, "setup_complete": false, "setup_id": "bc80a87d2dc174f9635cb42de3d06da8", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Depth prose measures speed, not complete answers. hidden facts JSON retrieval is finite evidence; synthesis semantics are unassessed. Whole-host memory includes unrelated services. Managed lab evidence, not customer acceptance." ], "engine": { "name": "Mirai S CUDA", "recipe": "with multi-token prediction", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "13f45764dce5469788295663", "measured_at": "2026-10-09T14:17:42.152672+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 36.05715063205489 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 7.61454219603911 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 12.519657135009766 } ], "model": { "artifact_sha256": [ "af3e2d508cb12ebf68688a0756da6bcb4cdfceca3883f5751666d5e4fb34a334" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "Mirai S trellis with Q8_0 MTP GGUF variant", "weights_dtype": "Mirai S trellis with Q8_0 MTP", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "1/1" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "q8_0", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "505b5b19fe93831f9fa0e211c935e6cb", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GeForce RTX 3060, 12 GB VRAM", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Speed only; outputs reach the token ceiling and do not establish correctness. Whole-Mac memory includes existing applications. Filled contexts observed whole-Mac swap growth; a memory-stable recipe, managed lifecycle and delivery remain unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "168b7cff6c2548c6942087f1", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 28.364087811849625 }, { "basis": null, "n": null, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 0.29438474999915343 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 56.456619224524324 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.16283139566803584 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 13.367050170898438 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 2, "context_tokens": 4096, "kv_dtype": "q8_0", "max_output_tokens": 512, "thinking": false }, "setup_complete": false, "setup_id": "bc80a87d2dc174f9635cb42de3d06da8", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance." ], "engine": { "name": "Mirai S CUDA", "recipe": "plain", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "1d3b3864eeed452b9756596f", "measured_at": "2026-10-09T17:00:14.157340+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 92.40260674903402 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20211 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 40.06080375099555 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 32.52640853391771 }, { "basis": null, "n": null, "name": "cache_hit_rate", "scope": "per_person", "source_field": "rolling_context_turns[0].cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40340 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 49.26079917198513 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 29.483310247922606 }, { "basis": null, "n": null, "name": "cache_hit_rate", "scope": "per_person", "source_field": "rolling_context_turns[1].cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5029251363411007 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 60470 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 62.26363143202616 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 17.6669226852687 }, { "basis": null, "n": null, "name": "cache_hit_rate", "scope": "per_person", "source_field": "rolling_context_turns[2].cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.668381015379527 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40513 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 92.40260674903402 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 30.75318701985158 }, { "basis": null, "n": null, "name": "cache_hit_rate", "scope": "per_person", "source_field": "rolling_context_turns[3].cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.003110112803297707 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 92.40260674903402 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 62.26363143202616 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 49.26079917198513 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 4 } ], "model": { "artifact_sha256": [ "af3e2d508cb12ebf68688a0756da6bcb4cdfceca3883f5751666d5e4fb34a334" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "MIRAI-S-TRELLIS-Q8-MTP variant", "weights_dtype": "MIRAI-S-TRELLIS-Q8-MTP", "weights_format": "gguf" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "4/4" } ], "reasons": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 64000, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "456ccf6622c78cec75a0a40e6b7ef10f", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GeForce RTX 3060, 12 GB VRAM", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Speed only; outputs reach the token ceiling and do not establish correctness. Whole-Mac memory includes existing applications. Filled contexts observed whole-Mac swap growth; a memory-stable recipe, managed lifecycle and delivery remain unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "21a91b3062bf439f871434e3", "measured_at": "2026-10-09T20:18:54.084891+00:00", "metrics": [ { "basis": null, "n": null, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 6.498728957987623 }, { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 25.07732863570589 }, { "basis": null, "n": null, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 6.602943457997753 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 25.07732863570589 }, { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "single", "unit": "tok/s", "value": 25.07732863570589 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 6.529574221994456 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 14.340469360351562 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "3/3" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 5120, "kv_dtype": "q8_0", "max_output_tokens": 512, "thinking": false }, "setup_complete": false, "setup_id": "2bcb8afd454b8f9078b745fcb7ebe3cd", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Four failed supplied-context answers are retained. No retrieval or permission test. Manifest is operator evidence; no customer or managed-service qualification.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "256c90ad21904342b440bec8", "measured_at": "2026-10-09T19:58:59.861332+00:00", "metrics": [ { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[0].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[0].passed", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[0].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "ttft_p95_s", "scope": "per_person", "source_field": "workload_levels[0].ttft_p95_s", "statistic": "p95", "unit": "s", "value": 0.3314811250020284 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[1].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[1].passed", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[1].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "ttft_p95_s", "scope": "per_person", "source_field": "workload_levels[1].ttft_p95_s", "statistic": "p95", "unit": "s", "value": 0.29090749999159016 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": false, "name": "structured-tasks", "observed": "28/32" } ], "reasons": [ "28/32 tasks passed." ], "status": "partial" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 28, "skipped": null, "suite": "structured-tasks", "total": 32, "version": null } ], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": { "concurrency": 2, "context_tokens": 4096, "kv_dtype": "q8_0", "max_output_tokens": 512, "thinking": false }, "setup_complete": false, "setup_id": "b29a0a490f1f9a21b71bb8031334432c", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Speed only; outputs reach the token ceiling and do not establish correctness. Whole-Mac memory includes existing applications. Filled contexts observed whole-Mac swap growth; a memory-stable recipe, managed lifecycle and delivery remain unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "38924c758a9544a8ab8924f0", "measured_at": "2026-10-09T20:23:27.615471+00:00", "metrics": [ { "basis": null, "n": null, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 11.270005895996292 }, { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 21.37946613294433 }, { "basis": null, "n": null, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 13.242090625004494 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 38.823101270046244 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 11.279436104164537 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 14.340469360351562 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 2, "context_tokens": 5120, "kv_dtype": "q8_0", "max_output_tokens": 512, "thinking": false }, "setup_complete": false, "setup_id": "2bcb8afd454b8f9078b745fcb7ebe3cd", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Finite non-refusal/refusal labels are not general safety certification. Truncated benign answers show non-refusal only." ], "engine": { "name": "Mirai S CUDA", "recipe": "with multi-token prediction", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "424b33c36c884565a9c6fc0a", "measured_at": "2026-10-09T14:13:52.607969+00:00", "metrics": [ { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 8 } ], "model": { "artifact_sha256": [ "af3e2d508cb12ebf68688a0756da6bcb4cdfceca3883f5751666d5e4fb34a334" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "Mirai S trellis with Q8_0 MTP GGUF variant", "weights_dtype": "Mirai S trellis with Q8_0 MTP", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "8/8" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": "93b444104682a88ecb100c566125d325362711858498e853994240a89049dd8d", "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": { "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "q8_0", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "52c42e50188aab94487646001da970af", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GeForce RTX 3060, 12 GB VRAM", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Depth prose measures speed, not complete answers. hidden facts JSON retrieval is finite evidence; synthesis semantics are unassessed. Whole-host memory includes unrelated services. Managed lab evidence, not customer acceptance." ], "engine": { "name": "Mirai S CUDA", "recipe": "with multi-token prediction", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "4357e9e5cef3429f8bf02334", "measured_at": "2026-10-09T14:26:28.539741+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 17.756948651371122 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 147.3930680649937 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 12.802810668945312 } ], "model": { "artifact_sha256": [ "af3e2d508cb12ebf68688a0756da6bcb4cdfceca3883f5751666d5e4fb34a334" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "Mirai S trellis with Q8_0 MTP GGUF variant", "weights_dtype": "Mirai S trellis with Q8_0 MTP", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "1/1" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "q8_0", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "b2c6ddc3fbc2f76fe56028793c8849f1", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GeForce RTX 3060, 12 GB VRAM", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Depth prose measures speed, not complete answers. hidden facts JSON retrieval is finite evidence; synthesis semantics are unassessed. Whole-host memory includes unrelated services. Managed lab evidence, not customer acceptance." ], "engine": { "name": "Mirai S CUDA", "recipe": "with multi-token prediction", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "61def83799144a138c96e3bc", "measured_at": "2026-10-09T14:37:25.662804+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 6.045408512411609 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 437.85434256098233 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 12.81662368774414 } ], "model": { "artifact_sha256": [ "af3e2d508cb12ebf68688a0756da6bcb4cdfceca3883f5751666d5e4fb34a334" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "Mirai S trellis with Q8_0 MTP GGUF variant", "weights_dtype": "Mirai S trellis with Q8_0 MTP", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "1/1" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "q8_0", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "a224df8dc4a1bb21465d808798760cf6", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GeForce RTX 3060, 12 GB VRAM", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Incomplete cohort; two-user filled-context speed not assessed in this attempt. Existing applications preserved. Shutdown during active work remains unqualified.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "6525551960bc44b2b0ea1bc6", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 14.712509155273438 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [], "reasons": [ "Incomplete cohort; two-user filled-context speed not assessed in this attempt." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": null, "context_tokens": 18432, "kv_dtype": "q8_0", "thinking": false }, "setup_complete": false, "setup_id": "9eb400231b01aaf6a0d2dd42c4de9ae1", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Depth prose measures speed, not complete answers. hidden facts JSON retrieval is finite evidence; synthesis semantics are unassessed. Whole-host memory includes unrelated services. Managed lab evidence, not customer acceptance." ], "engine": { "name": "Mirai S CUDA", "recipe": "with multi-token prediction", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "68a30d35b8064ec9b775f2d0", "measured_at": "2026-10-09T14:21:18.048632+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 33.076304527604655 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 66.86870725196786 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 12.584564208984375 } ], "model": { "artifact_sha256": [ "af3e2d508cb12ebf68688a0756da6bcb4cdfceca3883f5751666d5e4fb34a334" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "Mirai S trellis with Q8_0 MTP GGUF variant", "weights_dtype": "Mirai S trellis with Q8_0 MTP", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "1/1" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "q8_0", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "c87da726794d43b075eb72801c3350ca", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GeForce RTX 3060, 12 GB VRAM", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Retrieval completion and the 300-second first-token bar are independent. Synthesis semantics, broader quality and customer acceptance are unassessed." ], "engine": { "name": "Mirai S CUDA", "recipe": "with multi-token prediction", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "6a38761f11ed4674a966f55c", "measured_at": "2026-10-09T15:28:53.860792+00:00", "metrics": [ { "basis": null, "n": 1, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 1270.4283582619973 }, { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": 1, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 1270.4283582619973 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "long_context_requests[0].ttft_s", "statistic": "single", "unit": "s", "value": 1270.4283582619973 }, { "basis": null, "n": null, "name": "total_s", "scope": "per_person", "source_field": "long_context_requests[0].total_s", "statistic": "single", "unit": "s", "value": 1311.4375258379732 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "long_context_requests[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 2.7067118539861896 }, { "basis": "effective, includes queueing", "n": null, "name": "prefill_tps", "scope": "per_person", "source_field": "long_context_requests[0].effective_prefill_tps", "statistic": "single", "unit": "tok/s", "value": 196.79267892132555 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "long_context_requests[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "long_context_requests[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 250011 }, { "basis": null, "n": null, "name": "output_tokens", "scope": "per_person", "source_field": "long_context_requests[0].output_tokens", "statistic": "single", "unit": "tokens", "value": 112 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "per_person", "source_field": "long_context_requests[0].needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "needle_total", "scope": "per_person", "source_field": "long_context_requests[0].needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "first_real_token_from_dispatch_s", "scope": "per_person", "source_field": "long_context_requests[0].first_real_token_from_dispatch_s", "statistic": "single", "unit": "s", "value": 1270.4276904180297 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 12.85345458984375 } ], "model": { "artifact_sha256": [ "af3e2d508cb12ebf68688a0756da6bcb4cdfceca3883f5751666d5e4fb34a334" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "Mirai S trellis with Q8_0 MTP GGUF variant", "weights_dtype": "Mirai S trellis with Q8_0 MTP", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "1/1" }, { "met": true, "name": "hidden-facts", "observed": "3/3" }, { "met": false, "name": "first-token-300-s-reference", "observed": 1270.4283582619973 } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 3, "skipped": null, "suite": "hidden-facts", "total": 3, "version": null } ], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [ { "cached_prompt_tokens": null, "checks_passed": true, "completed": true, "decode_tps": 2.7067118539861896, "effective_prefill_tps": 196.79267892132555, "first_real_token_from_dispatch_s": 1270.4276904180297, "first_token_within_300_seconds": false, "needle_correct": 3, "needle_total": 3, "output_tokens": 112, "prompt_tokens": 250011, "stop_and_done": true, "strict_json": true, "total_s": 1311.4375258379732, "ttft_s": 1270.4283582619973, "user_number": 0, "valid_json": true } ], "scenario": { "id": "long-context-concurrency", "version": null }, "settings": { "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 250000, "fixture_variant": "original-tokenizer-needles-v1", "kv_dtype": "q8_0", "max_output_tokens": 512, "scenario": "distinct", "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "e9598834858cb992c23de5c7b679c340", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GeForce RTX 3060, 12 GB VRAM", "os": null }, "workload": { "cache": "cold", "content_tokens": 250000, "document": "distinct", "input_tokens": { "max": 250011, "median": 250011, "min": 250011 }, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Speed smoke does not grade task correctness. Sampled whole-host unavailable memory is not allocator residency. Configured context is not long-context quality proof.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Mirai S CUDA", "recipe": "with multi-token prediction", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "6c11f8fa03bf41feb5e151f4", "measured_at": "2026-10-09T13:54:43.418286+00:00", "metrics": [ { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 37.29158758439987 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_ttft_mean_s", "scope": "per_person", "source_field": "code_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.3206991030019708 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 41.474609113921616 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.3018271363495539 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 12.611061096191406 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "Mirai S trellis with Q8_0 MTP GGUF variant", "weights_dtype": "Mirai S trellis with Q8_0 MTP", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 262144, "kv_dtype": "q8_0", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction" }, "setup_complete": false, "setup_id": "6014d88c3f7c08e1f3c51a164593d2c7", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GeForce RTX 3060, 12 GB VRAM", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Speed only; outputs reach the token ceiling and do not establish correctness. Whole-Mac memory includes existing applications. Filled contexts observed whole-Mac swap growth; a memory-stable recipe, managed lifecycle and delivery remain unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "703b0c0b5e5b4734b0836524", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 0.13354741600051057 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "code_ttft_mean_s", "scope": "per_person", "source_field": "code_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.13161645799603625 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 28.030501209115137 }, { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "single", "unit": "tok/s", "value": 28.030501209115137 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 28.030501209115137 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 13.367050170898438 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "3/3" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 4096, "kv_dtype": "q8_0", "max_output_tokens": 512, "thinking": false }, "setup_complete": false, "setup_id": "a23eda5d4e76dad4cf1be4d98870c7cb", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Finite non-refusal/refusal labels are not general safety certification. Truncated benign answers show non-refusal only." ], "engine": { "name": "Mirai S CUDA", "recipe": "with multi-token prediction", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "729aab162682460490a8da60", "measured_at": "2026-10-09T14:13:52.607969+00:00", "metrics": [ { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 8 } ], "model": { "artifact_sha256": [ "af3e2d508cb12ebf68688a0756da6bcb4cdfceca3883f5751666d5e4fb34a334" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "Mirai S trellis with Q8_0 MTP GGUF variant", "weights_dtype": "Mirai S trellis with Q8_0 MTP", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "8/8" }, { "met": true, "name": "structured-tasks", "observed": "8/8" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 8, "skipped": null, "suite": "structured-tasks", "total": 8, "version": null } ], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": "93b444104682a88ecb100c566125d325362711858498e853994240a89049dd8d", "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": { "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "q8_0", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "52c42e50188aab94487646001da970af", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GeForce RTX 3060, 12 GB VRAM", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Bounded fictional tasks only. Unsupported cases are separately counted, never passes. Original publisher name with variant; exact variant repository and revision retained privately. Not customer acceptance.", "Publisher revision was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Mirai S CUDA", "recipe": "with multi-token prediction", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "78abd8c8e6e94c80a00efa84", "measured_at": "2026-10-09T14:00:47.877196+00:00", "metrics": [ { "basis": null, "n": null, "name": "skipped_cases", "scope": "machine_total", "source_field": "skipped_cases", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "correctness_failures", "scope": "machine_total", "source_field": "correctness_failures", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 0.6377832209691405 }, { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 122 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 119 }, { "basis": null, "n": null, "name": "api_cases_passed", "scope": "machine_total", "source_field": "api_cases_passed", "statistic": "single", "unit": "count", "value": 117 }, { "basis": null, "n": null, "name": "ttft_p95_s", "scope": "per_person", "source_field": "ttft_p95_s", "statistic": "p95", "unit": "s", "value": 1.0989372279727831 }, { "basis": null, "n": null, "name": "completion_p95_s", "scope": "per_person", "source_field": "completion_p95_s", "statistic": "p95", "unit": "s", "value": 1.7970678089768626 }, { "basis": null, "n": null, "name": "api_cases_total", "scope": "machine_total", "source_field": "api_cases_total", "statistic": "single", "unit": "count", "value": 120 }, { "basis": null, "n": null, "name": "completion_median_s", "scope": "per_person", "source_field": "completion_median_s", "statistic": "median", "unit": "s", "value": 0.8913421609904617 }, { "basis": null, "n": null, "name": "completeness_failures", "scope": "machine_total", "source_field": "completeness_failures", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[0].passed", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[0].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[0].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[1].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[1].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[1].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[2].passed", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[2].total", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[2].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[3].passed", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[3].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[3].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[4].passed", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[4].total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[4].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[5].passed", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[5].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[5].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[6].passed", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[6].total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[6].skipped", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[7].passed", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[7].total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[7].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[8].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[8].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[8].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[9].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[9].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[9].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[10].passed", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[10].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[10].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[11].passed", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[11].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[11].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[12].passed", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[12].total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[12].skipped", "statistic": "single", "unit": "count", "value": 4 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": null, "variant": "Mirai S trellis with Q8_0 MTP GGUF variant", "weights_dtype": "Mirai S trellis with Q8_0 MTP", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": false, "name": "structured-tasks", "observed": "119/122" }, { "met": false, "name": "API-cases", "observed": "117/120" } ], "reasons": [ "119/122 tasks passed.", "117/120 API cases passed." ], "status": "partial" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 119, "skipped": null, "suite": "structured-tasks", "total": 122, "version": "core-v2" }, { "passed": 117, "skipped": 8, "suite": "API-cases", "total": 120, "version": "core-v2" } ], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": "0ad0b13a94ad74bcdefe99d12d796b175f56e5777d9642a1dd4c0f170d708398", "harness": { "name": "keep-api-suite-run-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": "core-v2", "workload_version": "2" }, "requests": [], "scenario": { "id": "api-suite", "version": "1" }, "settings": { "concurrency": 1, "context_tokens": 262144, "kv_dtype": "q8_0", "speculative_decoding": "with multi-token prediction" }, "setup_complete": false, "setup_id": "e27a9e4f9fbdde55d07ca9d31990318c", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GeForce RTX 3060, 12 GB VRAM", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Speed only; outputs reach the token ceiling and do not establish correctness. Whole-Mac memory includes existing applications. Filled contexts observed whole-Mac swap growth; a memory-stable recipe, managed lifecycle and delivery remain unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "79b4656ba3e54d90907867aa", "measured_at": "2026-10-09T20:02:33.688657+00:00", "metrics": [ { "basis": null, "n": null, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 0.139347958000144 }, { "basis": null, "n": null, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 0.13948416699713562 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_ttft_mean_s", "scope": "per_person", "source_field": "code_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.13920859716745326 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 56.70901529200259 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 28.354696122449532 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 13.367050170898438 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 2, "context_tokens": 4096, "kv_dtype": "q8_0", "max_output_tokens": 512, "thinking": false }, "setup_complete": false, "setup_id": "a23eda5d4e76dad4cf1be4d98870c7cb", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Speed only; outputs reach the token ceiling and do not establish correctness. Whole-Mac memory includes existing applications. Filled contexts observed whole-Mac swap growth; a memory-stable recipe, managed lifecycle and delivery remain unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "80c51fd823774ccd9fe20379", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 6.494908583001234 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "code_ttft_mean_s", "scope": "per_person", "source_field": "code_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 6.486602402335848 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 25.152054873890684 }, { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "single", "unit": "tok/s", "value": 25.152054873890684 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 25.152054873890684 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 14.340469360351562 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "3/3" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 5120, "kv_dtype": "q8_0", "max_output_tokens": 512, "thinking": false }, "setup_complete": false, "setup_id": "73683dfa8a33ff83c0e60931bb8e3820", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Speed only; outputs reach the token ceiling and do not establish correctness. Whole-Mac memory includes existing applications. Filled contexts observed whole-Mac swap growth; a memory-stable recipe, managed lifecycle and delivery remain unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "8351831f3766456fa3047e69", "measured_at": "2026-10-09T20:00:13.341305+00:00", "metrics": [ { "basis": null, "n": null, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 0.13070437499845866 }, { "basis": null, "n": null, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 0.13354741600051057 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "code_ttft_mean_s", "scope": "per_person", "source_field": "code_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.13161645799603625 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 28.030501209115137 }, { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "single", "unit": "tok/s", "value": 28.030501209115137 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 28.030501209115137 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 13.367050170898438 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "3/3" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 4096, "kv_dtype": "q8_0", "max_output_tokens": 512, "thinking": false }, "setup_complete": false, "setup_id": "a23eda5d4e76dad4cf1be4d98870c7cb", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Operator-supplied runtime manifest; the endpoint runner cannot attest hardware or model pins. Bounded task correctness is not customer acceptance.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Mirai S CUDA", "recipe": "with multi-token prediction", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "858a9f7677b74eb184c298f1", "measured_at": "2026-10-09T13:55:12.337873+00:00", "metrics": [ { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[0].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[0].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[0].total", "statistic": "single", "unit": "count", "value": 16 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "Mirai S trellis with Q8_0 MTP GGUF variant", "weights_dtype": "Mirai S trellis with Q8_0 MTP", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "structured-tasks", "observed": "16/16" } ], "reasons": [], "status": "pass" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 16, "skipped": null, "suite": "structured-tasks", "total": 16, "version": null } ], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": { "concurrency": 1, "context_tokens": 262144, "kv_dtype": "q8_0", "speculative_decoding": "with multi-token prediction" }, "setup_complete": false, "setup_id": "6014d88c3f7c08e1f3c51a164593d2c7", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GeForce RTX 3060, 12 GB VRAM", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Startup, inference completion, concurrent decode and retrieval correctness are separate gates.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Mirai S CUDA", "recipe": "plain", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "8fd6f651a14c40f7b6700dad", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 0 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "unrecorded variant", "weights_dtype": "unrecorded", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/0" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [ "Startup, inference completion, concurrent decode and retrieval correctness are separate gates." ], "status": "did_not_start" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "long-context-concurrency", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "5cb97bdfed3d2480021d755103bdcb36", "standard": "pre-standard", "system": { "driver": null, "hardware": "unrecorded hardware", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Four failed supplied-context answers are retained. No retrieval or permission test. Manifest is operator evidence; no customer or managed-service qualification.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "93c24d3394ea4bcf8f82d006", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 28 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[0].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[0].passed", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[0].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[1].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[1].passed", "statistic": "single", "unit": "count", "value": 14 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[1].total", "statistic": "single", "unit": "count", "value": 16 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": false, "name": "structured-tasks", "observed": "28/32" } ], "reasons": [ "28/32 tasks passed." ], "status": "partial" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 28, "skipped": null, "suite": "structured-tasks", "total": 32, "version": null } ], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": { "concurrency": 2, "context_tokens": 4096, "kv_dtype": "q8_0", "max_output_tokens": 512, "thinking": false }, "setup_complete": false, "setup_id": "b29a0a490f1f9a21b71bb8031334432c", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. No summaries. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance." ], "engine": { "name": "Mirai S CUDA", "recipe": "plain", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "94cf38e463f04b01a0e25758", "measured_at": "2026-10-09T15:29:57.086713+00:00", "metrics": [ { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 4 } ], "model": { "artifact_sha256": [ "af3e2d508cb12ebf68688a0756da6bcb4cdfceca3883f5751666d5e4fb34a334" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "MIRAI-S-TRELLIS-Q8-MTP variant", "weights_dtype": "MIRAI-S-TRELLIS-Q8-MTP", "weights_format": "gguf" }, "outcome": { "gates": [ { "met": false, "name": "requests", "observed": "0/4" } ], "reasons": [ "0/4 requests passed." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 64000, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "66be08af32df5e60c9144779932efdb9", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GeForce RTX 3060, 12 GB VRAM", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Speed only; outputs reach the token ceiling and do not establish correctness. Whole-Mac memory includes existing applications. Filled contexts observed whole-Mac swap growth; a memory-stable recipe, managed lifecycle and delivery remain unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "9df3808e8f1542bba6652422", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 28.052985477395005 }, { "basis": null, "n": null, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 0.13460495800245553 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 28.052985477395005 }, { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "single", "unit": "tok/s", "value": 28.052985477395005 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.13200361066750096 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 13.367050170898438 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "3/3" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 4096, "kv_dtype": "q8_0", "max_output_tokens": 512, "thinking": false }, "setup_complete": false, "setup_id": "bc80a87d2dc174f9635cb42de3d06da8", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Speed only; outputs reach the token ceiling and do not establish correctness. Whole-Mac memory includes existing applications. Filled contexts observed whole-Mac swap growth; a memory-stable recipe, managed lifecycle and delivery remain unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "9f9d5dc50cf24a5ebf8803a3", "measured_at": "2026-10-09T20:01:21.062306+00:00", "metrics": [ { "basis": null, "n": null, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 0.13083916599862278 }, { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 28.052985477395005 }, { "basis": null, "n": null, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 0.13460495800245553 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 28.052985477395005 }, { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "single", "unit": "tok/s", "value": 28.052985477395005 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.13200361066750096 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 13.367050170898438 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "3/3" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 4096, "kv_dtype": "q8_0", "max_output_tokens": 512, "thinking": false }, "setup_complete": false, "setup_id": "bc80a87d2dc174f9635cb42de3d06da8", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance." ], "engine": { "name": "Mirai S CUDA", "recipe": "plain", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "b98a127d10c04a9e9ffc49b6", "measured_at": "2026-10-09T16:47:12.098204+00:00", "metrics": [ { "basis": null, "n": null, "name": "roll_ttft_max_seconds", "scope": "per_person", "source_field": "roll_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 91.11899668502156 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[0].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[0].turn_number", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 20211 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[0].ttft_s", "statistic": "single", "unit": "s", "value": 40.055931905983016 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[0].decode_tps", "statistic": "single", "unit": "tok/s", "value": 32.50707314283227 }, { "basis": null, "n": null, "name": "cache_hit_rate", "scope": "per_person", "source_field": "rolling_context_turns[0].cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.0 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[0].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[0].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[0].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[1].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[1].turn_number", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40340 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[1].ttft_s", "statistic": "single", "unit": "s", "value": 49.228578287002165 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[1].decode_tps", "statistic": "single", "unit": "tok/s", "value": 29.492327231949282 }, { "basis": null, "n": null, "name": "cache_hit_rate", "scope": "per_person", "source_field": "rolling_context_turns[1].cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.5029251363411007 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[1].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[1].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 40006 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[1].dropped_messages", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[2].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[2].turn_number", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40511 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[2].ttft_s", "statistic": "single", "unit": "s", "value": 91.04569341702154 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[2].decode_tps", "statistic": "single", "unit": "tok/s", "value": 32.25005258430812 }, { "basis": null, "n": null, "name": "cache_hit_rate", "scope": "per_person", "source_field": "rolling_context_turns[2].cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.003110266347411814 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[2].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[2].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 60009 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[2].dropped_messages", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "user_number", "scope": "per_person", "source_field": "rolling_context_turns[3].user_number", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "turn_number", "scope": "per_person", "source_field": "rolling_context_turns[3].turn_number", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "prompt_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].prompt_tokens", "statistic": "single", "unit": "tokens", "value": 40513 }, { "basis": null, "n": null, "name": "ttft_s", "scope": "per_person", "source_field": "rolling_context_turns[3].ttft_s", "statistic": "single", "unit": "s", "value": 91.11899668502156 }, { "basis": null, "n": null, "name": "decode_tps", "scope": "per_person", "source_field": "rolling_context_turns[3].decode_tps", "statistic": "single", "unit": "tok/s", "value": 30.832325882861387 }, { "basis": null, "n": null, "name": "cache_hit_rate", "scope": "per_person", "source_field": "rolling_context_turns[3].cache_hit_rate", "statistic": "single", "unit": "ratio", "value": 0.003110112803297707 }, { "basis": null, "n": null, "name": "content_tokens_added", "scope": "per_person", "source_field": "rolling_context_turns[3].content_tokens_added", "statistic": "single", "unit": "tokens", "value": 20003 }, { "basis": null, "n": null, "name": "cumulative_content_tokens", "scope": "per_person", "source_field": "rolling_context_turns[3].cumulative_content_tokens", "statistic": "single", "unit": "tokens", "value": 80012 }, { "basis": null, "n": null, "name": "dropped_messages", "scope": "per_person", "source_field": "rolling_context_turns[3].dropped_messages", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "roll_ttft_p50_seconds", "scope": "per_person", "source_field": "roll_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 91.08234505102155 }, { "basis": null, "n": null, "name": "rolled_off_abstained", "scope": "machine_total", "source_field": "rolled_off_abstained", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_invalid", "scope": "machine_total", "source_field": "rolled_off_invalid", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "rolled_off_wrong", "scope": "machine_total", "source_field": "rolled_off_wrong", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "normal_ttft_max_seconds", "scope": "per_person", "source_field": "normal_ttft_max_seconds", "statistic": "max", "unit": "s", "value": 49.228578287002165 }, { "basis": null, "n": null, "name": "rolled_off_correct", "scope": "machine_total", "source_field": "rolled_off_correct", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "normal_ttft_p50_seconds", "scope": "per_person", "source_field": "normal_ttft_p50_seconds", "statistic": "median", "unit": "s", "value": 44.64225509649259 }, { "basis": null, "n": null, "name": "rolled_off_total", "scope": "machine_total", "source_field": "rolled_off_total", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 4 } ], "model": { "artifact_sha256": [ "af3e2d508cb12ebf68688a0756da6bcb4cdfceca3883f5751666d5e4fb34a334" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "MIRAI-S-TRELLIS-Q8-MTP variant", "weights_dtype": "MIRAI-S-TRELLIS-Q8-MTP", "weights_format": "gguf" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "4/4" } ], "reasons": [ "Bounded tuning pilot; incomplete window growth cannot pass the full workload grade." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": "keep-rolling-context-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "rolling-context", "version": "1" }, "settings": { "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 64000, "max_output_tokens": 256 }, "setup_complete": false, "setup_id": "456ccf6622c78cec75a0a40e6b7ef10f", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GeForce RTX 3060, 12 GB VRAM", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 256, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Depth prose measures speed, not complete answers. hidden facts JSON retrieval is finite evidence; synthesis semantics are unassessed. Whole-host memory includes unrelated services. Managed lab evidence, not customer acceptance." ], "engine": { "name": "Mirai S CUDA", "recipe": "with multi-token prediction", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "bb974390fc3a47fb829b5c6d", "measured_at": "2026-10-09T15:03:47.502900+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 2.8812251954132293 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 1271.0217456920072 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 13.044918060302734 } ], "model": { "artifact_sha256": [ "af3e2d508cb12ebf68688a0756da6bcb4cdfceca3883f5751666d5e4fb34a334" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "Mirai S trellis with Q8_0 MTP GGUF variant", "weights_dtype": "Mirai S trellis with Q8_0 MTP", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "1/1" }, { "met": false, "name": "hidden-facts", "observed": "0/0" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 0, "skipped": null, "suite": "hidden-facts", "total": 0, "version": null } ], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "q8_0", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "6ea2aa234281a39f008d34c62b8ff22b", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GeForce RTX 3060, 12 GB VRAM", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Speed only; outputs reach the token ceiling and do not establish correctness. Whole-Mac memory includes existing applications. Filled contexts observed whole-Mac swap growth; a memory-stable recipe, managed lifecycle and delivery remain unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "bce93e95f8f5435aba8cac42", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 0.13948416699713562 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_ttft_mean_s", "scope": "per_person", "source_field": "code_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 0.13920859716745326 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 56.70901529200259 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 28.354696122449532 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 13.367050170898438 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 2, "context_tokens": 4096, "kv_dtype": "q8_0", "max_output_tokens": 512, "thinking": false }, "setup_complete": false, "setup_id": "a23eda5d4e76dad4cf1be4d98870c7cb", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Incomplete cohort; two-user filled-context speed not assessed in this attempt. Existing applications preserved. Shutdown during active work remains unqualified.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "bd0a65e6b9c940f2bf691afe", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 14.437286376953125 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [], "reasons": [ "Incomplete cohort; two-user filled-context speed not assessed in this attempt." ], "status": "fail" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": null, "context_tokens": 9216, "kv_dtype": "q8_0", "thinking": false }, "setup_complete": false, "setup_id": "41628bbb3739426de0fc069ee47a4313", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Speed only; outputs reach the token ceiling and do not establish correctness. Whole-Mac memory includes existing applications. Filled contexts observed whole-Mac swap growth; a memory-stable recipe, managed lifecycle and delivery remain unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "c57033fd02ed4319b1c821fd", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 21.37946613294433 }, { "basis": null, "n": null, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 13.242090625004494 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 38.823101270046244 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 11.279436104164537 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 14.340469360351562 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 2, "context_tokens": 5120, "kv_dtype": "q8_0", "max_output_tokens": 512, "thinking": false }, "setup_complete": false, "setup_id": "2bcb8afd454b8f9078b745fcb7ebe3cd", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Depth prose measures speed, not complete answers. hidden facts JSON retrieval is finite evidence; synthesis semantics are unassessed. Whole-host memory includes unrelated services. Managed lab evidence, not customer acceptance." ], "engine": { "name": "Mirai S CUDA", "recipe": "with multi-token prediction", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "cc19d195af9b405b94be15f3", "measured_at": "2026-10-09T15:28:53.799633+00:00", "metrics": [ { "basis": null, "n": null, "name": "needle_total", "scope": "machine_total", "source_field": "needle_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 2.7067118539861896 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "needle_correct", "scope": "machine_total", "source_field": "needle_correct", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "latency_bar_seconds", "scope": "machine_total", "source_field": "latency_bar_seconds", "statistic": "single", "unit": "s", "value": 300 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 1270.4283582619973 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 12.85345458984375 } ], "model": { "artifact_sha256": [ "af3e2d508cb12ebf68688a0756da6bcb4cdfceca3883f5751666d5e4fb34a334" ], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "variant": "Mirai S trellis with Q8_0 MTP GGUF variant", "weights_dtype": "Mirai S trellis with Q8_0 MTP", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "1/1" }, { "met": true, "name": "hidden-facts", "observed": "3/3" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 3, "skipped": null, "suite": "hidden-facts", "total": 3, "version": null } ], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 262144, "kv_dtype": "q8_0", "max_output_tokens": 512, "speculative_decoding": "with multi-token prediction", "thinking": false }, "setup_complete": false, "setup_id": "bb142c76320338be2d260775edd265df", "standard": "pre-standard", "system": { "driver": null, "hardware": "NVIDIA GeForce RTX 3060, 12 GB VRAM", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Speed only; outputs reach the token ceiling and do not establish correctness. Whole-Mac memory includes existing applications. Filled contexts observed whole-Mac swap growth; a memory-stable recipe, managed lifecycle and delivery remain unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "dc71448776e34a9489fd39ac", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "chat_decode_tps", "scope": "per_person", "source_field": "chat_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 25.07732863570589 }, { "basis": null, "n": null, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 6.602943457997753 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 25.07732863570589 }, { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "single", "unit": "tok/s", "value": 25.07732863570589 }, { "basis": null, "n": null, "name": "chat_ttft_mean_s", "scope": "per_person", "source_field": "chat_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 6.529574221994456 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 14.340469360351562 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "3/3" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 5120, "kv_dtype": "q8_0", "max_output_tokens": 512, "thinking": false }, "setup_complete": false, "setup_id": "2bcb8afd454b8f9078b745fcb7ebe3cd", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Speed only; outputs reach the token ceiling and do not establish correctness. Whole-Mac memory includes existing applications. Filled contexts observed whole-Mac swap growth; a memory-stable recipe, managed lifecycle and delivery remain unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "ecf9e0bea39a4c86b02050ff", "measured_at": "2026-10-09T20:21:14.805682+00:00", "metrics": [ { "basis": null, "n": null, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 11.263202041991462 }, { "basis": null, "n": null, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 13.17789899998752 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_ttft_mean_s", "scope": "per_person", "source_field": "code_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 11.275430992992673 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 39.61382340727263 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 21.520283224364476 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 14.340469360351562 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 2, "context_tokens": 5120, "kv_dtype": "q8_0", "max_output_tokens": 512, "thinking": false }, "setup_complete": false, "setup_id": "73683dfa8a33ff83c0e60931bb8e3820", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Speed only; outputs reach the token ceiling and do not establish correctness. Whole-Mac memory includes existing applications. Filled contexts observed whole-Mac swap growth; a memory-stable recipe, managed lifecycle and delivery remain unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "fe834f7c2a1c4c3eb44b85b9", "measured_at": "2026-10-09T20:17:13.812368+00:00", "metrics": [ { "basis": null, "n": null, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 6.490284540996072 }, { "basis": null, "n": null, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 6.494908583001234 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "code_ttft_mean_s", "scope": "per_person", "source_field": "code_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 6.486602402335848 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 25.152054873890684 }, { "basis": null, "n": null, "name": "single_user_decode_tps", "scope": "per_person", "source_field": "single_user_decode_tps", "statistic": "single", "unit": "tok/s", "value": 25.152054873890684 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 25.152054873890684 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 14.340469360351562 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "3/3" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 1, "context_tokens": 5120, "kv_dtype": "q8_0", "max_output_tokens": 512, "thinking": false }, "setup_complete": false, "setup_id": "73683dfa8a33ff83c0e60931bb8e3820", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 1, "warmup": null } }, { "caveats": [ "Speed only; outputs reach the token ceiling and do not establish correctness. Whole-Mac memory includes existing applications. Filled contexts observed whole-Mac swap growth; a memory-stable recipe, managed lifecycle and delivery remain unassessed.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "llama.cpp Metal", "recipe": "plain", "version": "b11182" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "ffdb1b9b8daa487a9999ca5b", "measured_at": null, "metrics": [ { "basis": null, "n": null, "name": "ttft_max_s", "scope": "per_person", "source_field": "ttft_max_s", "statistic": "max", "unit": "s", "value": 13.17789899998752 }, { "basis": null, "n": null, "name": "requests_ok", "scope": "machine_total", "source_field": "requests_ok", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "code_ttft_mean_s", "scope": "per_person", "source_field": "code_ttft_mean_s", "statistic": "mean", "unit": "s", "value": 11.275430992992673 }, { "basis": null, "n": null, "name": "aggregate_concurrency", "scope": "machine_total", "source_field": "aggregate_concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "requests_total", "scope": "machine_total", "source_field": "requests_total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "aggregate_e2e_tps", "scope": "machine_total", "source_field": "aggregate_tps", "statistic": "single", "unit": "tok/s", "value": 39.61382340727263 }, { "basis": null, "n": null, "name": "code_decode_tps", "scope": "per_person", "source_field": "code_decode_tps", "statistic": "mean", "unit": "tok/s", "value": 21.520283224364476 }, { "basis": null, "n": null, "name": "peak_host_unavailable_gib", "scope": "machine_total", "source_field": "peak_host_unavailable_gib", "statistic": "max", "unit": "GiB", "value": 14.340469360351562 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3-8B-Base", "publisher_revision": "49e3418fbbbca6ecbdf9608b4d22e5a407081db4", "variant": "Q4_K_M GGUF variant", "weights_dtype": "Q4_K_M", "weights_format": "GGUF" }, "outcome": { "gates": [ { "met": true, "name": "requests", "observed": "6/6" } ], "reasons": [], "status": "not_graded" }, "publication_group": "pre-standard", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-09", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "llm-throughput", "version": null }, "settings": { "concurrency": 2, "context_tokens": 5120, "kv_dtype": "q8_0", "max_output_tokens": 512, "thinking": false }, "setup_complete": false, "setup_id": "73683dfa8a33ff83c0e60931bb8e3820", "standard": "pre-standard", "system": { "driver": null, "hardware": "Apple M6 Mac mini, 16 GiB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 512, "repetitions": null, "users": 2, "warmup": null } }, { "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "Mirai S CUDA", "recipe": "plain", "version": "0.2.0-dev-vector-disabled" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "0388c0a8d393422585997e9e", "licence_class": "commercial", "measured_at": "2026-10-10T20:27:37.404250+00:00", "metrics": [ { "basis": null, "n": null, "name": "cold_decode_tps", "scope": "per_person", "source_field": "standard_result.cold_decode_tps", "statistic": "single", "unit": "tok/s", "value": 39.34606482087799 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.8-27B", "publisher_revision": null, "variant": "Mirai S trellis with Q8_0 MTP variant", "weights_dtype": "Mirai S trellis with Q8_0 MTP", "weights_format": "unrecorded" }, "outcome": { "gates": [], "reasons": [ "Standard check failed; the retained internal record contains the failure reason." ], "status": "fail" }, "publication_group": "what-did-not-work", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-10", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "aca52d36998a6522e509121270500cd0", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GeForce RTX 3060, 12 GB VRAM", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "bf13ac82f9944651b180398a", "licence_class": "commercial", "measured_at": "2026-10-10T20:30:57.890273+00:00", "metrics": [ { "basis": null, "n": null, "name": "cold_decode_tps", "scope": "per_person", "source_field": "standard_result.cold_decode_tps", "statistic": "single", "unit": "tok/s", "value": 104.02360247040544 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "unrecorded" }, "outcome": { "gates": [], "reasons": [ "Standard check failed; the retained internal record contains the failure reason." ], "status": "fail" }, "publication_group": "what-did-not-work", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-10", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": {}, "setup_complete": false, "setup_id": "8ad8d8a2bd6af1ff2c8637fe20b5948f", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "677d13aa74c347cf97289531754407a3", "role": "primary", "sha256": "02924ded8ab65987943e22eccd909e874d9e405f52ec04d83a859c3ea33bd7dd", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "1x128000-cold", "cell_outcome_sha256": "9784f388caed9732751df6f5062c4cf3d48a28a564f7712e3c7e0a0af9711c25", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "12c3230cf88c41528361b392", "licence_class": "commercial", "measured_at": "2026-10-18T23:28:13.676812+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 73.45597012204085 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 80.73324134079192 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 80.73293013821556 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 53.267056089127436 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 53.267056089127436 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 80.73324134079192 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 63.41 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0010517718639663498 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 83668.33437447368 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 73.1140999229495 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 73.11405237279894 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 52.18903221003711 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 52.18903221003711 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 73.1140999229495 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 66.29499999999999 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0010859662111749415 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 91163.05735966568 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 73.45597012204085 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 73.45568722424528 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 52.25469744182192 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 52.25469744182192 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 73.45597012204085 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 67.44 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.001105380021997474 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 89561.95881042098 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[3].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 75.04103802835652 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[3].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 75.04090171371612 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[3].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 52.308907526079565 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[3].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 52.308907526079565 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[3].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 75.04103802835652 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[3].power_w_median", "statistic": "median", "unit": "W", "value": 68.41499999999999 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[3].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0011186317647826466 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[3].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 88500.9733468779 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[4].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 71.95618472024205 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[4].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 71.95531055860816 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[4].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 52.31838228413835 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[4].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 52.31838228413835 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[4].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 71.95618472024205 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[4].power_w_median", "statistic": "median", "unit": "W", "value": 68.62 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[4].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0011249499393046116 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[4].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 78225.70314053018 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 128000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "adaptive_cell": { "attempts": [], "cell_id": "20x64000-cold", "cell_outcome_sha256": "99ee85ea758ec33172f659713c8f2d2e6dc85f8a7daf8721e9ad66b001419ccb", "cell_status": "above-confirmed-limit", "limit_reference": { "attempt_sha256": [ "7d93186c6fe2cf486ca8e2d248c0fca064137f0280d73630293467f41cae0765", "804b4cd3684668ae12890d6c8c09c755e344205da585720291303a836fb9b79f" ], "cell_id": "16x64000-cold", "payload_sha256": "e3baafba6bdabbde2673d14ab0973413053860df8f06c4b488d6a28d55b36ec5" }, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "21fc3056ebef4ba584b084bf", "licence_class": "commercial", "measured_at": "2026-10-18T23:28:13.066380+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "cold", "concurrency": 20, "configured_context_tokens": 262144, "context_tokens": 64000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 20, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "7e8579233dcf43159e704992785c36be", "role": "primary", "sha256": "d8d81fefb1b7f137e295f10be96dc2e203daa6eaf00f9701275a18cc28b0dd02", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "16x4000-cold", "cell_outcome_sha256": "1c086e2ce452326fd8ae2c5fa893bbb8a4cea0a5d14d343e644afac1eb85c442", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "31685df5df17425ebbfbac18", "licence_class": "commercial", "measured_at": "2026-10-18T19:52:14.885253+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 23.50188474669863 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 20.534125815280294 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 166.52467796989507 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 9.024816378485411 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 11.10994930495508 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 12.337601471855605 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 10.84655799722965 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 12.628627020047434 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 12.48229118854067 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 20.183475500237343 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 20.41783006607375 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 21.516092089211167 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 20.650421564486837 }, { "basis": null, "n": null, "name": "per_person_decode_tps[8]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[8]", "statistic": "single", "unit": "tok/s", "value": 13.016324275494872 }, { "basis": null, "n": null, "name": "per_person_decode_tps[9]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[9]", "statistic": "single", "unit": "tok/s", "value": 21.759395091708413 }, { "basis": null, "n": null, "name": "per_person_decode_tps[10]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[10]", "statistic": "single", "unit": "tok/s", "value": 37.114179775214346 }, { "basis": null, "n": null, "name": "per_person_decode_tps[11]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[11]", "statistic": "single", "unit": "tok/s", "value": 27.952776228812432 }, { "basis": null, "n": null, "name": "per_person_decode_tps[12]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[12]", "statistic": "single", "unit": "tok/s", "value": 13.673172009153687 }, { "basis": null, "n": null, "name": "per_person_decode_tps[13]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[13]", "statistic": "single", "unit": "tok/s", "value": 23.6566955559026 }, { "basis": null, "n": null, "name": "per_person_decode_tps[14]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[14]", "statistic": "single", "unit": "tok/s", "value": 38.72315924634627 }, { "basis": null, "n": null, "name": "per_person_decode_tps[15]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[15]", "statistic": "single", "unit": "tok/s", "value": 38.90071815542247 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[8]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[8]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[9]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[9]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[10]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[10]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[11]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[11]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[12]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[12]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[13]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[13]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[14]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[14]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[15]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[15]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 12.135000000000002 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0002367666771071937 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 6166408.287847871 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 22.819777989634076 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 185.91743903176393 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 7.9212261030916125 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 9.608287723036483 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 13.970837418399983 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 12.11435147161216 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 22.507272765708553 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 14.616350265794823 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 26.89811964787043 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 22.685272252535576 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 14.818330218792106 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 26.78511828323101 }, { "basis": null, "n": null, "name": "per_person_decode_tps[8]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[8]", "statistic": "single", "unit": "tok/s", "value": 27.402692652743532 }, { "basis": null, "n": null, "name": "per_person_decode_tps[9]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[9]", "statistic": "single", "unit": "tok/s", "value": 32.24183888242257 }, { "basis": null, "n": null, "name": "per_person_decode_tps[10]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[10]", "statistic": "single", "unit": "tok/s", "value": 37.05847938994207 }, { "basis": null, "n": null, "name": "per_person_decode_tps[11]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[11]", "statistic": "single", "unit": "tok/s", "value": 22.95428372673258 }, { "basis": null, "n": null, "name": "per_person_decode_tps[12]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[12]", "statistic": "single", "unit": "tok/s", "value": 15.182022921427134 }, { "basis": null, "n": null, "name": "per_person_decode_tps[13]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[13]", "statistic": "single", "unit": "tok/s", "value": 36.89019520145855 }, { "basis": null, "n": null, "name": "per_person_decode_tps[14]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[14]", "statistic": "single", "unit": "tok/s", "value": 15.929509270076359 }, { "basis": null, "n": null, "name": "per_person_decode_tps[15]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[15]", "statistic": "single", "unit": "tok/s", "value": 38.42364085564877 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[8]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[8]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[9]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[9]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[10]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[10]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[11]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[11]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[12]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[12]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[13]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[13]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[14]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[14]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[15]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[15]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 12.29 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00023394410280700573 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 6232257.973191101 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 23.97432198857603 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 187.39906464528045 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 7.9160931130172685 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 9.607378646032885 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 13.943192385084794 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 12.04870763887641 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 14.358942988237695 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 14.408036604063838 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 14.413349030964406 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 27.216104695313504 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 26.883541438543638 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 23.227110246904843 }, { "basis": null, "n": null, "name": "per_person_decode_tps[8]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[8]", "statistic": "single", "unit": "tok/s", "value": 32.370085063993265 }, { "basis": null, "n": null, "name": "per_person_decode_tps[9]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[9]", "statistic": "single", "unit": "tok/s", "value": 37.20926568465786 }, { "basis": null, "n": null, "name": "per_person_decode_tps[10]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[10]", "statistic": "single", "unit": "tok/s", "value": 35.86960467817089 }, { "basis": null, "n": null, "name": "per_person_decode_tps[11]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[11]", "statistic": "single", "unit": "tok/s", "value": 23.81415845729382 }, { "basis": null, "n": null, "name": "per_person_decode_tps[12]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[12]", "statistic": "single", "unit": "tok/s", "value": 15.348146670937533 }, { "basis": null, "n": null, "name": "per_person_decode_tps[13]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[13]", "statistic": "single", "unit": "tok/s", "value": 29.008293227822087 }, { "basis": null, "n": null, "name": "per_person_decode_tps[14]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[14]", "statistic": "single", "unit": "tok/s", "value": 24.13448551985824 }, { "basis": null, "n": null, "name": "per_person_decode_tps[15]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[15]", "statistic": "single", "unit": "tok/s", "value": 38.519135835676394 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[8]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[8]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[9]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[9]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[10]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[10]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[11]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[11]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[12]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[12]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[13]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[13]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[14]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[14]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[15]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[15]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 12.39 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00023531293726515495 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 6247000.343816953 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[3].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 23.8948388256941 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[3].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 185.2528874729564 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[3].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 7.908293351531029 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[3].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 9.608528094831854 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[3].decode_overlap", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 14.070410180140883 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 11.857989871147014 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 14.064326767646815 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 14.272969041212368 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 26.690422362921005 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 26.587504535550927 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 27.118871014641552 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 22.92631155523631 }, { "basis": null, "n": null, "name": "per_person_decode_tps[8]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[8]", "statistic": "single", "unit": "tok/s", "value": 31.86848876941895 }, { "basis": null, "n": null, "name": "per_person_decode_tps[9]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[9]", "statistic": "single", "unit": "tok/s", "value": 36.57810169254562 }, { "basis": null, "n": null, "name": "per_person_decode_tps[10]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[10]", "statistic": "single", "unit": "tok/s", "value": 15.002046595566142 }, { "basis": null, "n": null, "name": "per_person_decode_tps[11]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[11]", "statistic": "single", "unit": "tok/s", "value": 36.79115955584791 }, { "basis": null, "n": null, "name": "per_person_decode_tps[12]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[12]", "statistic": "single", "unit": "tok/s", "value": 15.457713102484336 }, { "basis": null, "n": null, "name": "per_person_decode_tps[13]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[13]", "statistic": "single", "unit": "tok/s", "value": 23.817750540372217 }, { "basis": null, "n": null, "name": "per_person_decode_tps[14]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[14]", "statistic": "single", "unit": "tok/s", "value": 23.971927111015987 }, { "basis": null, "n": null, "name": "per_person_decode_tps[15]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[15]", "statistic": "single", "unit": "tok/s", "value": 37.18955307762884 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[8]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[8]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[9]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[9]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[10]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[10]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[11]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[11]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[12]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[12]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[13]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[13]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[14]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[14]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[15]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[15]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[3].power_w_median", "statistic": "median", "unit": "W", "value": 12.45 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[3].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00023665532151859598 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[3].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 6211565.370967117 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[4].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 23.50188474669863 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[4].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 185.5404124561376 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[4].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 7.920604337588884 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[4].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 9.612591393990442 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[4].decode_overlap", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 13.91374461616418 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 13.951381336694439 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 14.109144510884501 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 12.063845377915126 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 14.36965418482816 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 22.32004251849588 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 14.199425231134171 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 22.57341068751887 }, { "basis": null, "n": null, "name": "per_person_decode_tps[8]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[8]", "statistic": "single", "unit": "tok/s", "value": 27.080288236055345 }, { "basis": null, "n": null, "name": "per_person_decode_tps[9]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[9]", "statistic": "single", "unit": "tok/s", "value": 32.19107326741416 }, { "basis": null, "n": null, "name": "per_person_decode_tps[10]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[10]", "statistic": "single", "unit": "tok/s", "value": 27.7188034289327 }, { "basis": null, "n": null, "name": "per_person_decode_tps[11]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[11]", "statistic": "single", "unit": "tok/s", "value": 37.47888691536329 }, { "basis": null, "n": null, "name": "per_person_decode_tps[12]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[12]", "statistic": "single", "unit": "tok/s", "value": 37.33055357433669 }, { "basis": null, "n": null, "name": "per_person_decode_tps[13]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[13]", "statistic": "single", "unit": "tok/s", "value": 24.430358805878384 }, { "basis": null, "n": null, "name": "per_person_decode_tps[14]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[14]", "statistic": "single", "unit": "tok/s", "value": 28.142156682486707 }, { "basis": null, "n": null, "name": "per_person_decode_tps[15]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[15]", "statistic": "single", "unit": "tok/s", "value": 39.04644126778932 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[8]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[8]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[9]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[9]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[10]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[10]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[11]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[11]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[12]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[12]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[13]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[13]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[14]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[14]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[15]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[15]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[4].power_w_median", "statistic": "median", "unit": "W", "value": 12.5 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[4].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00023716632976332784 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[4].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 6134934.927111991 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 16, "configured_context_tokens": 262144, "context_tokens": 4000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 16, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "c6593799d53a4b829fa5a0151b5962d2", "role": "primary", "sha256": "cf7f47233f25dbfe3e57774a09aa28702971b8e9c1acb2ee93559db4f6bad692", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "20x4000-cold", "cell_outcome_sha256": "f68f7ec8e42d7205f6df1736cab481083956768fcfdf49ef9efdc5e68d0ea0b2", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "3a1e6bf7154c4850b4f62cbf", "licence_class": "commercial", "measured_at": "2026-10-18T19:59:31.423046+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 16.739288450248768 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 16.352866930676434 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 168.75051771430213 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 9.898999951547012 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 12.779891291866079 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 9.423988758761407 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 9.57678058134791 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 9.737330792113543 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 8.791637803356505 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 10.045870992603627 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 16.105431354888555 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 13.58406291942785 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 10.202009655199419 }, { "basis": null, "n": null, "name": "per_person_decode_tps[8]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[8]", "statistic": "single", "unit": "tok/s", "value": 22.77603330607418 }, { "basis": null, "n": null, "name": "per_person_decode_tps[9]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[9]", "statistic": "single", "unit": "tok/s", "value": 13.831712843259037 }, { "basis": null, "n": null, "name": "per_person_decode_tps[10]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[10]", "statistic": "single", "unit": "tok/s", "value": 13.88096953529286 }, { "basis": null, "n": null, "name": "per_person_decode_tps[11]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[11]", "statistic": "single", "unit": "tok/s", "value": 25.629732757791285 }, { "basis": null, "n": null, "name": "per_person_decode_tps[12]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[12]", "statistic": "single", "unit": "tok/s", "value": 16.600302506464313 }, { "basis": null, "n": null, "name": "per_person_decode_tps[13]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[13]", "statistic": "single", "unit": "tok/s", "value": 32.17063815293925 }, { "basis": null, "n": null, "name": "per_person_decode_tps[14]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[14]", "statistic": "single", "unit": "tok/s", "value": 32.86466492190242 }, { "basis": null, "n": null, "name": "per_person_decode_tps[15]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[15]", "statistic": "single", "unit": "tok/s", "value": 32.88273659926958 }, { "basis": null, "n": null, "name": "per_person_decode_tps[16]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[16]", "statistic": "single", "unit": "tok/s", "value": 32.13321127699621 }, { "basis": null, "n": null, "name": "per_person_decode_tps[17]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[17]", "statistic": "single", "unit": "tok/s", "value": 33.166618361940124 }, { "basis": null, "n": null, "name": "per_person_decode_tps[18]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[18]", "statistic": "single", "unit": "tok/s", "value": 17.125085124995042 }, { "basis": null, "n": null, "name": "per_person_decode_tps[19]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[19]", "statistic": "single", "unit": "tok/s", "value": 33.01623507538772 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[8]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[8]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[9]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[9]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[10]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[10]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[11]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[11]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[12]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[12]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[13]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[13]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[14]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[14]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[15]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[15]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[16]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[16]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[17]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[17]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[18]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[18]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[19]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[19]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 12.26 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00027245181578989657 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 6654387.656561297 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 16.739288450248768 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 168.74863584720111 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 9.440420411527157 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 12.1624243860133 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 9.736236871477896 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 13.14335575788187 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 13.480167261273719 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 10.151883629305214 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 10.1961594226833 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 13.508697939678612 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 10.192323982421305 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 16.92698052453263 }, { "basis": null, "n": null, "name": "per_person_decode_tps[8]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[8]", "statistic": "single", "unit": "tok/s", "value": 16.551596375964905 }, { "basis": null, "n": null, "name": "per_person_decode_tps[9]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[9]", "statistic": "single", "unit": "tok/s", "value": 8.970690302625988 }, { "basis": null, "n": null, "name": "per_person_decode_tps[10]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[10]", "statistic": "single", "unit": "tok/s", "value": 21.057074105825034 }, { "basis": null, "n": null, "name": "per_person_decode_tps[11]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[11]", "statistic": "single", "unit": "tok/s", "value": 17.20665432320074 }, { "basis": null, "n": null, "name": "per_person_decode_tps[12]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[12]", "statistic": "single", "unit": "tok/s", "value": 31.17088254574912 }, { "basis": null, "n": null, "name": "per_person_decode_tps[13]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[13]", "statistic": "single", "unit": "tok/s", "value": 31.19169174902622 }, { "basis": null, "n": null, "name": "per_person_decode_tps[14]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[14]", "statistic": "single", "unit": "tok/s", "value": 31.475982288297995 }, { "basis": null, "n": null, "name": "per_person_decode_tps[15]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[15]", "statistic": "single", "unit": "tok/s", "value": 31.93316291184161 }, { "basis": null, "n": null, "name": "per_person_decode_tps[16]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[16]", "statistic": "single", "unit": "tok/s", "value": 10.913789741593842 }, { "basis": null, "n": null, "name": "per_person_decode_tps[17]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[17]", "statistic": "single", "unit": "tok/s", "value": 27.002444965431394 }, { "basis": null, "n": null, "name": "per_person_decode_tps[18]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[18]", "statistic": "single", "unit": "tok/s", "value": 33.11838043262551 }, { "basis": null, "n": null, "name": "per_person_decode_tps[19]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[19]", "statistic": "single", "unit": "tok/s", "value": 33.905579280822494 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[8]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[8]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[9]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[9]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[10]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[10]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[11]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[11]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[12]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[12]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[13]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[13]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[14]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[14]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[15]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[15]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[16]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[16]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[17]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[17]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[18]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[18]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[19]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[19]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 12.38 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0002752174846346044 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 6609318.453785798 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 16.980835831202505 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 172.06578563666358 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 9.382113119936548 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 12.105922101996839 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 20 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 9.734022415991223 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 9.89065379632492 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 9.888493788872497 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 13.159313197195399 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 10.052303369976768 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 9.01419218919831 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 10.265765349770868 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 13.744240100429598 }, { "basis": null, "n": null, "name": "per_person_decode_tps[8]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[8]", "statistic": "single", "unit": "tok/s", "value": 16.663644455594845 }, { "basis": null, "n": null, "name": "per_person_decode_tps[9]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[9]", "statistic": "single", "unit": "tok/s", "value": 20.97005461009817 }, { "basis": null, "n": null, "name": "per_person_decode_tps[10]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[10]", "statistic": "single", "unit": "tok/s", "value": 31.72730429784688 }, { "basis": null, "n": null, "name": "per_person_decode_tps[11]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[11]", "statistic": "single", "unit": "tok/s", "value": 17.29802720681017 }, { "basis": null, "n": null, "name": "per_person_decode_tps[12]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[12]", "statistic": "single", "unit": "tok/s", "value": 26.912849079889984 }, { "basis": null, "n": null, "name": "per_person_decode_tps[13]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[13]", "statistic": "single", "unit": "tok/s", "value": 32.97773750251896 }, { "basis": null, "n": null, "name": "per_person_decode_tps[14]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[14]", "statistic": "single", "unit": "tok/s", "value": 15.167566952457719 }, { "basis": null, "n": null, "name": "per_person_decode_tps[15]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[15]", "statistic": "single", "unit": "tok/s", "value": 32.918669929346464 }, { "basis": null, "n": null, "name": "per_person_decode_tps[16]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[16]", "statistic": "single", "unit": "tok/s", "value": 32.52308411181182 }, { "basis": null, "n": null, "name": "per_person_decode_tps[17]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[17]", "statistic": "single", "unit": "tok/s", "value": 18.49289916667578 }, { "basis": null, "n": null, "name": "per_person_decode_tps[18]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[18]", "statistic": "single", "unit": "tok/s", "value": 32.65666173637294 }, { "basis": null, "n": null, "name": "per_person_decode_tps[19]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[19]", "statistic": "single", "unit": "tok/s", "value": 32.955927732636894 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[8]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[8]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[9]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[9]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[10]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[10]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[11]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[11]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[12]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[12]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[13]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[13]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[14]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[14]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[15]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[15]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[16]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[16]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[17]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[17]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[18]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[18]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[19]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[19]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 12.59 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00027553396215612366 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 6667054.7094267635 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 20, "configured_context_tokens": 262144, "context_tokens": 4000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 20, "warmup": null } }, { "adaptive_cell": {}, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "3cbd32287f804c53abeb4143", "licence_class": "commercial", "measured_at": "2026-10-18T19:23:28.454945+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "cold_decode_tps", "scope": "per_person", "source_field": "standard_result.cold_decode_tps", "statistic": "single", "unit": "tok/s", "value": 105.56279928316732 }, { "basis": null, "n": null, "name": "idle_power_w", "scope": "machine_total", "source_field": "standard_result.idle_power_w", "statistic": "single", "unit": "W", "value": 5.2 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "configured_context_tokens": 262144 }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": "observed-serialization-boundary", "classification": "capacity", "id": "f585b9a1ee7b4934b9386ba935d35c57", "role": "primary", "sha256": "84ac08b120b0dadcf1b45cfcca0cfd8abab1c71ac1a416dea0a94a31b4c46fb4", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, { "capacity_kind": null, "classification": "measured", "id": "84dab8bacde749c5af9d8b2e8a934935", "role": "confirmation", "sha256": "e16cc71812487be8206974999142c12e6c19bdfdb89b28aac1681027dece45d4", "variant_sha256": "ce7a11417103cc42fa9c8145854c46461d126b984c768acf8e243600a467a158" } ], "cell_id": "64x4000-cold", "cell_outcome_sha256": "d4436f6ffd490fefea0921fa550fae9bfff3daeccd3618541a6051ec97bb940a", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "ce7a11417103cc42fa9c8145854c46461d126b984c768acf8e243600a467a158" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "4301ebe8ec824fb18f5a4d1d", "licence_class": "commercial", "measured_at": "2026-10-18T20:26:01.269182+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 4.295539634038278 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 4.197266390343523 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 141.82122435405046 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 23.2997450720286 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 40.66743550589308 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 64 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 2.2872401993227043 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 2.353277966801052 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 2.481615289329579 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 2.4814244987255116 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 2.48147179945199 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 2.3936835609095106 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 2.4175166148268605 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 2.5183277645348725 }, { "basis": null, "n": null, "name": "per_person_decode_tps[8]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[8]", "statistic": "single", "unit": "tok/s", "value": 2.600988102546482 }, { "basis": null, "n": null, "name": "per_person_decode_tps[9]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[9]", "statistic": "single", "unit": "tok/s", "value": 2.7870801934346803 }, { "basis": null, "n": null, "name": "per_person_decode_tps[10]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[10]", "statistic": "single", "unit": "tok/s", "value": 2.6681427249085568 }, { "basis": null, "n": null, "name": "per_person_decode_tps[11]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[11]", "statistic": "single", "unit": "tok/s", "value": 2.6615734496446875 }, { "basis": null, "n": null, "name": "per_person_decode_tps[12]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[12]", "statistic": "single", "unit": "tok/s", "value": 2.7631270878645036 }, { "basis": null, "n": null, "name": "per_person_decode_tps[13]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[13]", "statistic": "single", "unit": "tok/s", "value": 2.899044911353252 }, { "basis": null, "n": null, "name": "per_person_decode_tps[14]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[14]", "statistic": "single", "unit": "tok/s", "value": 2.6458287588939076 }, { "basis": null, "n": null, "name": "per_person_decode_tps[15]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[15]", "statistic": "single", "unit": "tok/s", "value": 2.751753085268637 }, { "basis": null, "n": null, "name": "per_person_decode_tps[16]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[16]", "statistic": "single", "unit": "tok/s", "value": 3.0296572815876917 }, { "basis": null, "n": null, "name": "per_person_decode_tps[17]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[17]", "statistic": "single", "unit": "tok/s", "value": 3.121321580273238 }, { "basis": null, "n": null, "name": "per_person_decode_tps[18]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[18]", "statistic": "single", "unit": "tok/s", "value": 3.2585905955167247 }, { "basis": null, "n": null, "name": "per_person_decode_tps[19]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[19]", "statistic": "single", "unit": "tok/s", "value": 3.0307696033986855 }, { "basis": null, "n": null, "name": "per_person_decode_tps[20]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[20]", "statistic": "single", "unit": "tok/s", "value": 3.143953017057804 }, { "basis": null, "n": null, "name": "per_person_decode_tps[21]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[21]", "statistic": "single", "unit": "tok/s", "value": 3.557267095838858 }, { "basis": null, "n": null, "name": "per_person_decode_tps[22]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[22]", "statistic": "single", "unit": "tok/s", "value": 3.4652785903989716 }, { "basis": null, "n": null, "name": "per_person_decode_tps[23]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[23]", "statistic": "single", "unit": "tok/s", "value": 3.5817269170479697 }, { "basis": null, "n": null, "name": "per_person_decode_tps[24]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[24]", "statistic": "single", "unit": "tok/s", "value": 3.622978241727008 }, { "basis": null, "n": null, "name": "per_person_decode_tps[25]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[25]", "statistic": "single", "unit": "tok/s", "value": 3.5280463723217004 }, { "basis": null, "n": null, "name": "per_person_decode_tps[26]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[26]", "statistic": "single", "unit": "tok/s", "value": 3.553394083002512 }, { "basis": null, "n": null, "name": "per_person_decode_tps[27]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[27]", "statistic": "single", "unit": "tok/s", "value": 3.8238872600930867 }, { "basis": null, "n": null, "name": "per_person_decode_tps[28]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[28]", "statistic": "single", "unit": "tok/s", "value": 3.974706681222053 }, { "basis": null, "n": null, "name": "per_person_decode_tps[29]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[29]", "statistic": "single", "unit": "tok/s", "value": 4.366824831995346 }, { "basis": null, "n": null, "name": "per_person_decode_tps[30]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[30]", "statistic": "single", "unit": "tok/s", "value": 4.160206618549571 }, { "basis": null, "n": null, "name": "per_person_decode_tps[31]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[31]", "statistic": "single", "unit": "tok/s", "value": 4.234326162137474 }, { "basis": null, "n": null, "name": "per_person_decode_tps[32]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[32]", "statistic": "single", "unit": "tok/s", "value": 3.5236585289392797 }, { "basis": null, "n": null, "name": "per_person_decode_tps[33]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[33]", "statistic": "single", "unit": "tok/s", "value": 4.090452198783132 }, { "basis": null, "n": null, "name": "per_person_decode_tps[34]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[34]", "statistic": "single", "unit": "tok/s", "value": 4.8513906813828065 }, { "basis": null, "n": null, "name": "per_person_decode_tps[35]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[35]", "statistic": "single", "unit": "tok/s", "value": 5.288199178113605 }, { "basis": null, "n": null, "name": "per_person_decode_tps[36]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[36]", "statistic": "single", "unit": "tok/s", "value": 5.356028106980931 }, { "basis": null, "n": null, "name": "per_person_decode_tps[37]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[37]", "statistic": "single", "unit": "tok/s", "value": 5.151764817597785 }, { "basis": null, "n": null, "name": "per_person_decode_tps[38]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[38]", "statistic": "single", "unit": "tok/s", "value": 5.000712742328665 }, { "basis": null, "n": null, "name": "per_person_decode_tps[39]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[39]", "statistic": "single", "unit": "tok/s", "value": 5.511359497671672 }, { "basis": null, "n": null, "name": "per_person_decode_tps[40]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[40]", "statistic": "single", "unit": "tok/s", "value": 5.891139319778349 }, { "basis": null, "n": null, "name": "per_person_decode_tps[41]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[41]", "statistic": "single", "unit": "tok/s", "value": 5.9212452804497495 }, { "basis": null, "n": null, "name": "per_person_decode_tps[42]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[42]", "statistic": "single", "unit": "tok/s", "value": 4.825154547450755 }, { "basis": null, "n": null, "name": "per_person_decode_tps[43]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[43]", "statistic": "single", "unit": "tok/s", "value": 7.317466898339487 }, { "basis": null, "n": null, "name": "per_person_decode_tps[44]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[44]", "statistic": "single", "unit": "tok/s", "value": 9.043606005520928 }, { "basis": null, "n": null, "name": "per_person_decode_tps[45]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[45]", "statistic": "single", "unit": "tok/s", "value": 6.485249597558 }, { "basis": null, "n": null, "name": "per_person_decode_tps[46]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[46]", "statistic": "single", "unit": "tok/s", "value": 8.144950587060134 }, { "basis": null, "n": null, "name": "per_person_decode_tps[47]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[47]", "statistic": "single", "unit": "tok/s", "value": 7.531182179273058 }, { "basis": null, "n": null, "name": "per_person_decode_tps[48]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[48]", "statistic": "single", "unit": "tok/s", "value": 8.279653839192777 }, { "basis": null, "n": null, "name": "per_person_decode_tps[49]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[49]", "statistic": "single", "unit": "tok/s", "value": 10.517315609781381 }, { "basis": null, "n": null, "name": "per_person_decode_tps[50]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[50]", "statistic": "single", "unit": "tok/s", "value": 6.917794378126092 }, { "basis": null, "n": null, "name": "per_person_decode_tps[51]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[51]", "statistic": "single", "unit": "tok/s", "value": 7.789662666422984 }, { "basis": null, "n": null, "name": "per_person_decode_tps[52]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[52]", "statistic": "single", "unit": "tok/s", "value": 5.002489867912057 }, { "basis": null, "n": null, "name": "per_person_decode_tps[53]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[53]", "statistic": "single", "unit": "tok/s", "value": 12.33793016907898 }, { "basis": null, "n": null, "name": "per_person_decode_tps[54]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[54]", "statistic": "single", "unit": "tok/s", "value": 10.929732978012929 }, { "basis": null, "n": null, "name": "per_person_decode_tps[55]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[55]", "statistic": "single", "unit": "tok/s", "value": 7.419683914349719 }, { "basis": null, "n": null, "name": "per_person_decode_tps[56]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[56]", "statistic": "single", "unit": "tok/s", "value": 12.546581934334066 }, { "basis": null, "n": null, "name": "per_person_decode_tps[57]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[57]", "statistic": "single", "unit": "tok/s", "value": 18.856474937641448 }, { "basis": null, "n": null, "name": "per_person_decode_tps[58]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[58]", "statistic": "single", "unit": "tok/s", "value": 18.850523935371097 }, { "basis": null, "n": null, "name": "per_person_decode_tps[59]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[59]", "statistic": "single", "unit": "tok/s", "value": 16.49945258887697 }, { "basis": null, "n": null, "name": "per_person_decode_tps[60]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[60]", "statistic": "single", "unit": "tok/s", "value": 13.546914087625806 }, { "basis": null, "n": null, "name": "per_person_decode_tps[61]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[61]", "statistic": "single", "unit": "tok/s", "value": 19.351634162406587 }, { "basis": null, "n": null, "name": "per_person_decode_tps[62]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[62]", "statistic": "single", "unit": "tok/s", "value": 16.08382680867115 }, { "basis": null, "n": null, "name": "per_person_decode_tps[63]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[63]", "statistic": "single", "unit": "tok/s", "value": 16.759238478646182 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[8]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[8]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[9]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[9]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[10]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[10]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[11]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[11]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[12]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[12]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[13]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[13]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[14]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[14]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[15]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[15]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[16]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[16]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[17]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[17]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[18]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[18]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[19]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[19]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[20]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[20]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[21]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[21]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[22]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[22]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[23]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[23]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[24]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[24]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[25]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[25]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[26]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[26]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[27]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[27]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[28]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[28]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[29]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[29]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[30]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[30]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[31]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[31]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[32]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[32]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[33]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[33]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[34]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[34]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[35]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[35]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[36]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[36]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[37]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[37]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[38]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[38]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[39]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[39]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[40]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[40]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[41]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[41]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[42]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[42]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[43]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[43]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[44]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[44]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[45]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[45]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[46]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[46]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[47]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[47]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[48]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[48]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[49]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[49]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[50]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[50]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[51]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[51]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[52]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[52]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[53]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[53]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[54]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[54]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[55]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[55]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[56]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[56]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[57]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[57]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[58]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[58]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[59]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[59]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[60]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[60]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[61]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[61]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[62]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[62]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[63]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[63]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 12.94 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0007888514015869422 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 7338517.73395369 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 4.295539634038278 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 145.74418361983828 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 22.48294685361907 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 38.9896264069248 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 64 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 2.3670664978758125 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 2.4849250751669705 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 2.5032859411814727 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 2.644133891044603 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 2.7187332748399173 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 2.5775425361698696 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 2.783686036429421 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 2.693090037501519 }, { "basis": null, "n": null, "name": "per_person_decode_tps[8]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[8]", "statistic": "single", "unit": "tok/s", "value": 2.6998466418725804 }, { "basis": null, "n": null, "name": "per_person_decode_tps[9]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[9]", "statistic": "single", "unit": "tok/s", "value": 2.9238910850775786 }, { "basis": null, "n": null, "name": "per_person_decode_tps[10]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[10]", "statistic": "single", "unit": "tok/s", "value": 2.7500144643135647 }, { "basis": null, "n": null, "name": "per_person_decode_tps[11]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[11]", "statistic": "single", "unit": "tok/s", "value": 2.92372402827303 }, { "basis": null, "n": null, "name": "per_person_decode_tps[12]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[12]", "statistic": "single", "unit": "tok/s", "value": 2.7191999985556157 }, { "basis": null, "n": null, "name": "per_person_decode_tps[13]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[13]", "statistic": "single", "unit": "tok/s", "value": 2.6876824206998795 }, { "basis": null, "n": null, "name": "per_person_decode_tps[14]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[14]", "statistic": "single", "unit": "tok/s", "value": 3.0966173833174007 }, { "basis": null, "n": null, "name": "per_person_decode_tps[15]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[15]", "statistic": "single", "unit": "tok/s", "value": 2.780630355145546 }, { "basis": null, "n": null, "name": "per_person_decode_tps[16]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[16]", "statistic": "single", "unit": "tok/s", "value": 3.5444394318023114 }, { "basis": null, "n": null, "name": "per_person_decode_tps[17]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[17]", "statistic": "single", "unit": "tok/s", "value": 3.1135563341378694 }, { "basis": null, "n": null, "name": "per_person_decode_tps[18]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[18]", "statistic": "single", "unit": "tok/s", "value": 3.3857940110203537 }, { "basis": null, "n": null, "name": "per_person_decode_tps[19]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[19]", "statistic": "single", "unit": "tok/s", "value": 3.364296800001962 }, { "basis": null, "n": null, "name": "per_person_decode_tps[20]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[20]", "statistic": "single", "unit": "tok/s", "value": 2.888702975048273 }, { "basis": null, "n": null, "name": "per_person_decode_tps[21]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[21]", "statistic": "single", "unit": "tok/s", "value": 3.6459611309966533 }, { "basis": null, "n": null, "name": "per_person_decode_tps[22]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[22]", "statistic": "single", "unit": "tok/s", "value": 3.729671775452855 }, { "basis": null, "n": null, "name": "per_person_decode_tps[23]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[23]", "statistic": "single", "unit": "tok/s", "value": 3.406107404920114 }, { "basis": null, "n": null, "name": "per_person_decode_tps[24]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[24]", "statistic": "single", "unit": "tok/s", "value": 3.773482090033345 }, { "basis": null, "n": null, "name": "per_person_decode_tps[25]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[25]", "statistic": "single", "unit": "tok/s", "value": 4.24603236528661 }, { "basis": null, "n": null, "name": "per_person_decode_tps[26]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[26]", "statistic": "single", "unit": "tok/s", "value": 4.323005813317264 }, { "basis": null, "n": null, "name": "per_person_decode_tps[27]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[27]", "statistic": "single", "unit": "tok/s", "value": 3.450187864249756 }, { "basis": null, "n": null, "name": "per_person_decode_tps[28]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[28]", "statistic": "single", "unit": "tok/s", "value": 4.497604364747858 }, { "basis": null, "n": null, "name": "per_person_decode_tps[29]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[29]", "statistic": "single", "unit": "tok/s", "value": 4.1220563477320935 }, { "basis": null, "n": null, "name": "per_person_decode_tps[30]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[30]", "statistic": "single", "unit": "tok/s", "value": 4.268073454759292 }, { "basis": null, "n": null, "name": "per_person_decode_tps[31]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[31]", "statistic": "single", "unit": "tok/s", "value": 5.174136042465495 }, { "basis": null, "n": null, "name": "per_person_decode_tps[32]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[32]", "statistic": "single", "unit": "tok/s", "value": 4.712383157348267 }, { "basis": null, "n": null, "name": "per_person_decode_tps[33]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[33]", "statistic": "single", "unit": "tok/s", "value": 5.463954430887459 }, { "basis": null, "n": null, "name": "per_person_decode_tps[34]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[34]", "statistic": "single", "unit": "tok/s", "value": 4.194942069746641 }, { "basis": null, "n": null, "name": "per_person_decode_tps[35]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[35]", "statistic": "single", "unit": "tok/s", "value": 3.9101674454795123 }, { "basis": null, "n": null, "name": "per_person_decode_tps[36]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[36]", "statistic": "single", "unit": "tok/s", "value": 5.56063569485478 }, { "basis": null, "n": null, "name": "per_person_decode_tps[37]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[37]", "statistic": "single", "unit": "tok/s", "value": 5.658036965451687 }, { "basis": null, "n": null, "name": "per_person_decode_tps[38]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[38]", "statistic": "single", "unit": "tok/s", "value": 4.631806991068346 }, { "basis": null, "n": null, "name": "per_person_decode_tps[39]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[39]", "statistic": "single", "unit": "tok/s", "value": 5.2371383686856445 }, { "basis": null, "n": null, "name": "per_person_decode_tps[40]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[40]", "statistic": "single", "unit": "tok/s", "value": 4.201052181153473 }, { "basis": null, "n": null, "name": "per_person_decode_tps[41]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[41]", "statistic": "single", "unit": "tok/s", "value": 6.0789031680922045 }, { "basis": null, "n": null, "name": "per_person_decode_tps[42]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[42]", "statistic": "single", "unit": "tok/s", "value": 6.76767489423515 }, { "basis": null, "n": null, "name": "per_person_decode_tps[43]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[43]", "statistic": "single", "unit": "tok/s", "value": 7.221335138854553 }, { "basis": null, "n": null, "name": "per_person_decode_tps[44]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[44]", "statistic": "single", "unit": "tok/s", "value": 6.464183552340006 }, { "basis": null, "n": null, "name": "per_person_decode_tps[45]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[45]", "statistic": "single", "unit": "tok/s", "value": 7.4686767747576885 }, { "basis": null, "n": null, "name": "per_person_decode_tps[46]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[46]", "statistic": "single", "unit": "tok/s", "value": 5.375065430228376 }, { "basis": null, "n": null, "name": "per_person_decode_tps[47]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[47]", "statistic": "single", "unit": "tok/s", "value": 8.22918346914633 }, { "basis": null, "n": null, "name": "per_person_decode_tps[48]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[48]", "statistic": "single", "unit": "tok/s", "value": 10.089274352967712 }, { "basis": null, "n": null, "name": "per_person_decode_tps[49]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[49]", "statistic": "single", "unit": "tok/s", "value": 11.740236888755444 }, { "basis": null, "n": null, "name": "per_person_decode_tps[50]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[50]", "statistic": "single", "unit": "tok/s", "value": 8.186968617086364 }, { "basis": null, "n": null, "name": "per_person_decode_tps[51]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[51]", "statistic": "single", "unit": "tok/s", "value": 9.226398069876522 }, { "basis": null, "n": null, "name": "per_person_decode_tps[52]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[52]", "statistic": "single", "unit": "tok/s", "value": 9.381238224830515 }, { "basis": null, "n": null, "name": "per_person_decode_tps[53]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[53]", "statistic": "single", "unit": "tok/s", "value": 13.850335957226909 }, { "basis": null, "n": null, "name": "per_person_decode_tps[54]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[54]", "statistic": "single", "unit": "tok/s", "value": 8.649471681886638 }, { "basis": null, "n": null, "name": "per_person_decode_tps[55]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[55]", "statistic": "single", "unit": "tok/s", "value": 12.156440029265672 }, { "basis": null, "n": null, "name": "per_person_decode_tps[56]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[56]", "statistic": "single", "unit": "tok/s", "value": 15.41232321006067 }, { "basis": null, "n": null, "name": "per_person_decode_tps[57]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[57]", "statistic": "single", "unit": "tok/s", "value": 14.122771028745044 }, { "basis": null, "n": null, "name": "per_person_decode_tps[58]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[58]", "statistic": "single", "unit": "tok/s", "value": 18.225328989362147 }, { "basis": null, "n": null, "name": "per_person_decode_tps[59]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[59]", "statistic": "single", "unit": "tok/s", "value": 18.006497131300396 }, { "basis": null, "n": null, "name": "per_person_decode_tps[60]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[60]", "statistic": "single", "unit": "tok/s", "value": 14.579920229577976 }, { "basis": null, "n": null, "name": "per_person_decode_tps[61]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[61]", "statistic": "single", "unit": "tok/s", "value": 17.761834900347697 }, { "basis": null, "n": null, "name": "per_person_decode_tps[62]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[62]", "statistic": "single", "unit": "tok/s", "value": 19.18212179752573 }, { "basis": null, "n": null, "name": "per_person_decode_tps[63]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[63]", "statistic": "single", "unit": "tok/s", "value": 17.28048626059217 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[8]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[8]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[9]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[9]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[10]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[10]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[11]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[11]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[12]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[12]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[13]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[13]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[14]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[14]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[15]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[15]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[16]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[16]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[17]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[17]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[18]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[18]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[19]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[19]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[20]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[20]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[21]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[21]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[22]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[22]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[23]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[23]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[24]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[24]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[25]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[25]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[26]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[26]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[27]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[27]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[28]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[28]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[29]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[29]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[30]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[30]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[31]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[31]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[32]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[32]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[33]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[33]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[34]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[34]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[35]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[35]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[36]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[36]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[37]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[37]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[38]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[38]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[39]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[39]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[40]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[40]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[41]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[41]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[42]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[42]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[43]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[43]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[44]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[44]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[45]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[45]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[46]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[46]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[47]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[47]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[48]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[48]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[49]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[49]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[50]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[50]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[51]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[51]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[52]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[52]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[53]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[53]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[54]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[54]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[55]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[55]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[56]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[56]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[57]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[57]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[58]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[58]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[59]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[59]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[60]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[60]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[61]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[61]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[62]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[62]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[63]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[63]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 13.184999999999999 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0008148427242537588 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 7144937.086272502 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 4.503003099439376 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 145.7128102161564 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 22.504815625376068 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 39.00006235297769 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 64 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 2.384896286351399 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 2.4093091436284526 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 2.5028490639559164 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 2.502958809415715 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 2.6494558415862732 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 2.5775584624323202 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 2.663744774172699 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 2.7083033557765495 }, { "basis": null, "n": null, "name": "per_person_decode_tps[8]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[8]", "statistic": "single", "unit": "tok/s", "value": 2.633961823327136 }, { "basis": null, "n": null, "name": "per_person_decode_tps[9]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[9]", "statistic": "single", "unit": "tok/s", "value": 2.8691904618242585 }, { "basis": null, "n": null, "name": "per_person_decode_tps[10]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[10]", "statistic": "single", "unit": "tok/s", "value": 2.731745957471811 }, { "basis": null, "n": null, "name": "per_person_decode_tps[11]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[11]", "statistic": "single", "unit": "tok/s", "value": 2.693730712520503 }, { "basis": null, "n": null, "name": "per_person_decode_tps[12]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[12]", "statistic": "single", "unit": "tok/s", "value": 2.8734919839402737 }, { "basis": null, "n": null, "name": "per_person_decode_tps[13]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[13]", "statistic": "single", "unit": "tok/s", "value": 2.913163075717055 }, { "basis": null, "n": null, "name": "per_person_decode_tps[14]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[14]", "statistic": "single", "unit": "tok/s", "value": 2.771383705962447 }, { "basis": null, "n": null, "name": "per_person_decode_tps[15]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[15]", "statistic": "single", "unit": "tok/s", "value": 3.0987267855651814 }, { "basis": null, "n": null, "name": "per_person_decode_tps[16]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[16]", "statistic": "single", "unit": "tok/s", "value": 3.062562666583387 }, { "basis": null, "n": null, "name": "per_person_decode_tps[17]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[17]", "statistic": "single", "unit": "tok/s", "value": 3.0606210993967666 }, { "basis": null, "n": null, "name": "per_person_decode_tps[18]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[18]", "statistic": "single", "unit": "tok/s", "value": 2.743725421534058 }, { "basis": null, "n": null, "name": "per_person_decode_tps[19]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[19]", "statistic": "single", "unit": "tok/s", "value": 3.4901012220008085 }, { "basis": null, "n": null, "name": "per_person_decode_tps[20]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[20]", "statistic": "single", "unit": "tok/s", "value": 3.754449186371639 }, { "basis": null, "n": null, "name": "per_person_decode_tps[21]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[21]", "statistic": "single", "unit": "tok/s", "value": 3.4097340741824524 }, { "basis": null, "n": null, "name": "per_person_decode_tps[22]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[22]", "statistic": "single", "unit": "tok/s", "value": 3.736578016909224 }, { "basis": null, "n": null, "name": "per_person_decode_tps[23]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[23]", "statistic": "single", "unit": "tok/s", "value": 3.631843955227072 }, { "basis": null, "n": null, "name": "per_person_decode_tps[24]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[24]", "statistic": "single", "unit": "tok/s", "value": 3.7595053234114233 }, { "basis": null, "n": null, "name": "per_person_decode_tps[25]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[25]", "statistic": "single", "unit": "tok/s", "value": 4.558758916414814 }, { "basis": null, "n": null, "name": "per_person_decode_tps[26]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[26]", "statistic": "single", "unit": "tok/s", "value": 4.293719615239817 }, { "basis": null, "n": null, "name": "per_person_decode_tps[27]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[27]", "statistic": "single", "unit": "tok/s", "value": 5.375553877160101 }, { "basis": null, "n": null, "name": "per_person_decode_tps[28]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[28]", "statistic": "single", "unit": "tok/s", "value": 3.663501036192246 }, { "basis": null, "n": null, "name": "per_person_decode_tps[29]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[29]", "statistic": "single", "unit": "tok/s", "value": 4.483481892907283 }, { "basis": null, "n": null, "name": "per_person_decode_tps[30]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[30]", "statistic": "single", "unit": "tok/s", "value": 5.501373090732164 }, { "basis": null, "n": null, "name": "per_person_decode_tps[31]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[31]", "statistic": "single", "unit": "tok/s", "value": 4.155986293362123 }, { "basis": null, "n": null, "name": "per_person_decode_tps[32]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[32]", "statistic": "single", "unit": "tok/s", "value": 5.017643457453421 }, { "basis": null, "n": null, "name": "per_person_decode_tps[33]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[33]", "statistic": "single", "unit": "tok/s", "value": 4.798512219694598 }, { "basis": null, "n": null, "name": "per_person_decode_tps[34]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[34]", "statistic": "single", "unit": "tok/s", "value": 5.2718209191741146 }, { "basis": null, "n": null, "name": "per_person_decode_tps[35]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[35]", "statistic": "single", "unit": "tok/s", "value": 4.510252497098913 }, { "basis": null, "n": null, "name": "per_person_decode_tps[36]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[36]", "statistic": "single", "unit": "tok/s", "value": 4.030536061143153 }, { "basis": null, "n": null, "name": "per_person_decode_tps[37]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[37]", "statistic": "single", "unit": "tok/s", "value": 4.0272154834934035 }, { "basis": null, "n": null, "name": "per_person_decode_tps[38]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[38]", "statistic": "single", "unit": "tok/s", "value": 4.5888888588669055 }, { "basis": null, "n": null, "name": "per_person_decode_tps[39]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[39]", "statistic": "single", "unit": "tok/s", "value": 4.49575370177984 }, { "basis": null, "n": null, "name": "per_person_decode_tps[40]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[40]", "statistic": "single", "unit": "tok/s", "value": 6.545055931648947 }, { "basis": null, "n": null, "name": "per_person_decode_tps[41]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[41]", "statistic": "single", "unit": "tok/s", "value": 6.1159018971037025 }, { "basis": null, "n": null, "name": "per_person_decode_tps[42]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[42]", "statistic": "single", "unit": "tok/s", "value": 6.114927291548544 }, { "basis": null, "n": null, "name": "per_person_decode_tps[43]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[43]", "statistic": "single", "unit": "tok/s", "value": 5.795294753038368 }, { "basis": null, "n": null, "name": "per_person_decode_tps[44]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[44]", "statistic": "single", "unit": "tok/s", "value": 7.18626121157624 }, { "basis": null, "n": null, "name": "per_person_decode_tps[45]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[45]", "statistic": "single", "unit": "tok/s", "value": 7.972928190785608 }, { "basis": null, "n": null, "name": "per_person_decode_tps[46]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[46]", "statistic": "single", "unit": "tok/s", "value": 7.227360329897592 }, { "basis": null, "n": null, "name": "per_person_decode_tps[47]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[47]", "statistic": "single", "unit": "tok/s", "value": 7.9703342240508555 }, { "basis": null, "n": null, "name": "per_person_decode_tps[48]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[48]", "statistic": "single", "unit": "tok/s", "value": 10.079420941188792 }, { "basis": null, "n": null, "name": "per_person_decode_tps[49]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[49]", "statistic": "single", "unit": "tok/s", "value": 9.012050856465335 }, { "basis": null, "n": null, "name": "per_person_decode_tps[50]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[50]", "statistic": "single", "unit": "tok/s", "value": 13.726510454499973 }, { "basis": null, "n": null, "name": "per_person_decode_tps[51]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[51]", "statistic": "single", "unit": "tok/s", "value": 12.017539718639531 }, { "basis": null, "n": null, "name": "per_person_decode_tps[52]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[52]", "statistic": "single", "unit": "tok/s", "value": 15.243394965437393 }, { "basis": null, "n": null, "name": "per_person_decode_tps[53]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[53]", "statistic": "single", "unit": "tok/s", "value": 5.957358789417142 }, { "basis": null, "n": null, "name": "per_person_decode_tps[54]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[54]", "statistic": "single", "unit": "tok/s", "value": 18.143778415017294 }, { "basis": null, "n": null, "name": "per_person_decode_tps[55]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[55]", "statistic": "single", "unit": "tok/s", "value": 14.408773567715041 }, { "basis": null, "n": null, "name": "per_person_decode_tps[56]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[56]", "statistic": "single", "unit": "tok/s", "value": 14.268307831010892 }, { "basis": null, "n": null, "name": "per_person_decode_tps[57]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[57]", "statistic": "single", "unit": "tok/s", "value": 16.12095814846395 }, { "basis": null, "n": null, "name": "per_person_decode_tps[58]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[58]", "statistic": "single", "unit": "tok/s", "value": 17.88624641957679 }, { "basis": null, "n": null, "name": "per_person_decode_tps[59]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[59]", "statistic": "single", "unit": "tok/s", "value": 18.716408633191833 }, { "basis": null, "n": null, "name": "per_person_decode_tps[60]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[60]", "statistic": "single", "unit": "tok/s", "value": 9.087108339832998 }, { "basis": null, "n": null, "name": "per_person_decode_tps[61]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[61]", "statistic": "single", "unit": "tok/s", "value": 18.628645173081114 }, { "basis": null, "n": null, "name": "per_person_decode_tps[62]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[62]", "statistic": "single", "unit": "tok/s", "value": 9.799610940846858 }, { "basis": null, "n": null, "name": "per_person_decode_tps[63]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[63]", "statistic": "single", "unit": "tok/s", "value": 13.324352213511375 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[8]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[8]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[9]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[9]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[10]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[10]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[11]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[11]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[12]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[12]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[13]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[13]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[14]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[14]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[15]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[15]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[16]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[16]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[17]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[17]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[18]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[18]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[19]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[19]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[20]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[20]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[21]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[21]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[22]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[22]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[23]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[23]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[24]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[24]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[25]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[25]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[26]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[26]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[27]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[27]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[28]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[28]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[29]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[29]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[30]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[30]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[31]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[31]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[32]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[32]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[33]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[33]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[34]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[34]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[35]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[35]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[36]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[36]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[37]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[37]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[38]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[38]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[39]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[39]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[40]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[40]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[41]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[41]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[42]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[42]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[43]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[43]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[44]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[44]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[45]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[45]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[46]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[46]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[47]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[47]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[48]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[48]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[49]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[49]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[50]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[50]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[51]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[51]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[52]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[52]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[53]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[53]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[54]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[54]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[55]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[55]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[56]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[56]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[57]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[57]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[58]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[58]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[59]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[59]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[60]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[60]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[61]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[61]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[62]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[62]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[63]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[63]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 13.36 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0008242671881715812 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 7051111.682478087 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "ce7a11417103cc42fa9c8145854c46461d126b984c768acf8e243600a467a158", "cache_state": "cold", "concurrency": 64, "configured_context_tokens": 262144, "context_tokens": 4000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 64, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "ce929b6d54f247e8bbd55f772b2741cf", "role": "primary", "sha256": "d788b32cbd2e21767ae513c94ec793d4c675bbb2dd9a29ed787a06edf5d25f85", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "8x40000-cold", "cell_outcome_sha256": "9b28be1390b4677e66c216f53023f99e0a620e0489ba0b628e65412a7afaa4f3", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "45102e8a8b9e4c27aeb73141", "licence_class": "commercial", "measured_at": "2026-10-18T21:06:48.566704+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 7.617398603419042 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 7.617398603419042 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 53.157646101984085 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 56.083195076091215 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 66.95885404595174 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 7.193540692970418 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 7.154472165225402 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 7.4974643479635255 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 7.296435902238271 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 7.737332858874559 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 8.031788214617942 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 42.35846032070385 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 43.04716990018537 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 60.375 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0012434642486201066 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 601545.2401064753 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 7.353906277993667 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 47.399782861325626 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 56.086903170566075 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 66.93781173089519 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 7.063995861185454 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 7.144926880340017 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 7.277959698516188 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 7.600356368738026 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 7.429852857471145 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 6.715186025258956 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 44.0600945557004 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 47.367691348720136 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 63.03 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0012974178194978551 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 573446.7253486261 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 7.985634298179294 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 50.17595133566363 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 55.59389539505355 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 65.69505501189269 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 6.130019611887198 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 7.5270755371111795 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 7.668222278680943 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 7.714068813113625 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 8.299677531302919 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 8.257199783244962 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 42.47414976971776 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 45.10689573680269 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 65.69 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0013232899452969378 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 557700.9049474774 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 8, "configured_context_tokens": 262144, "context_tokens": 40000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 8, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "b4a216e00eda4a46ab11f5ca48c200cc", "role": "primary", "sha256": "c09dec415600b431ce80e61c902526a5f9e5b4ec745064331dee24ca1697d51e", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "5x40000-cold", "cell_outcome_sha256": "7e9705d9e6e9c55c1b3f235a3a2b30ae6b07a05884e0a62e314ec4c8b8d27551", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "4bf8d4f0df2d4dce828cbfd6", "licence_class": "commercial", "measured_at": "2026-10-18T20:59:08.748957+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 49.705583969246234 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 50.23998124188693 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 206.70483614447633 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 41.44161124411039 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 41.4453886509873 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 43.435727137176194 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 50.093664745945624 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 50.23998124188693 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 50.944594669280065 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 50.663124804005314 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 56.224999999999994 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0007915366785201199 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 587464.8801738127 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 49.705583969246234 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 140.16899163474974 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 41.20564030716196 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 41.207325944909826 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 29.296773707576627 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 49.705583969246234 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 49.99285969362027 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 49.37680286224928 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 50.45204203714992 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 58.81 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0008229216436749589 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 580857.2464631911 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 49.45158672383189 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 146.79481052093917 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 40.69612693786621 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 40.69706217292696 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 49.45158672383189 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 30.479822240822365 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 45.143502157110866 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 49.58037426412802 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 50.86498977462053 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 60.47 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0008366629966258311 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 545022.3110607226 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 40000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 5, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": "observed-throughput-boundary", "classification": "capacity", "id": "ade0e19db2cd4781a157f5fd0af7537c", "role": "primary", "sha256": "aee01025e9c568df2b02c38b20918b42eb99e0da86fd5111099cb8d7c9c5cb93", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, { "capacity_kind": "observed-throughput-boundary", "classification": "capacity", "id": "07ed60868174492d974f8e507a9b54c4", "role": "confirmation", "sha256": "777aff85ef232ae148a855cb769c47394647360313ff9e999ad3c71e99fb8638", "variant_sha256": "ce7a11417103cc42fa9c8145854c46461d126b984c768acf8e243600a467a158" } ], "cell_id": "16x40000-cold", "cell_outcome_sha256": "442b5f767aec4fb4d71a3c1945592a8efdbf5a7aa308a9f2e988d527b5eee9ae", "cell_status": "capacity-limit-confirmed", "limit_reference": null, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "609fad9f190846208161ad2b", "licence_class": "commercial", "measured_at": "2026-10-18T21:15:46.757249+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "cold", "concurrency": 16, "configured_context_tokens": 262144, "context_tokens": 40000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 16, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "a2127f13540f4dbc94fd934f28f5dbb8", "role": "primary", "sha256": "f17270e3b08facca571723785015e2230711ecb59c86de481d62e1f4c2991479", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "2x128000-cold", "cell_outcome_sha256": "cc8af62fb94aa3e495750416919682048e5460f5f544106ecd50051c3dc91671", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "647a66220fbb479780605021", "licence_class": "commercial", "measured_at": "2026-10-18T23:39:28.463093+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 45.2220822334669 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 59.3899526895411 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 116.72359831197689 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 93.37034447095357 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 93.39607455185615 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 59.75491615485541 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 59.024989224226786 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 71.525 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0019517833998725107 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 91198.64428175117 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 45.204069202151885 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 69.09078696126257 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 94.94653574796394 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 95.40507329604588 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 35.31600804175968 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 55.09213036254408 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 73.41 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0020443997388653154 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 89023.68579885155 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 45.2220822334669 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 69.72591201036201 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 94.12904667039402 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 94.59642868884839 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 35.71059163052865 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 54.73357283640515 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 74.37 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0020691482376325406 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 86025.73598287112 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[3].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 44.86983299711339 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[3].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 68.64178481890019 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[3].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 94.22582476504613 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[3].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 94.69346255809069 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[3].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 35.094344280356886 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 54.645321713869905 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[3].power_w_median", "statistic": "median", "unit": "W", "value": 74.69 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[3].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0020856259110904477 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[3].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 85346.08198597541 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[4].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 45.97640333997626 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[4].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 69.33525078965567 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[4].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 94.26454693509731 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[4].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 94.73527447506785 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[4].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 54.87511627418781 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 37.0776904057647 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[4].power_w_median", "statistic": "median", "unit": "W", "value": 75.29 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[4].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0020973341921521243 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[4].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 90114.3941233622 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 2, "configured_context_tokens": 262144, "context_tokens": 128000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 2, "warmup": null } }, { "adaptive_cell": { "attempts": [], "cell_id": "64x40000-cold", "cell_outcome_sha256": "0bc9d3971ccf6a66fb76fc07232387c6bb91406a900b783a164fa98219bb7745", "cell_status": "above-confirmed-limit", "limit_reference": { "attempt_sha256": [ "aee01025e9c568df2b02c38b20918b42eb99e0da86fd5111099cb8d7c9c5cb93", "777aff85ef232ae148a855cb769c47394647360313ff9e999ad3c71e99fb8638" ], "cell_id": "16x40000-cold", "payload_sha256": "442b5f767aec4fb4d71a3c1945592a8efdbf5a7aa308a9f2e988d527b5eee9ae" }, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "6a67e7b610cc4795a0aa1197", "licence_class": "commercial", "measured_at": "2026-10-18T21:44:27.452199+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "cold", "concurrency": 64, "configured_context_tokens": 262144, "context_tokens": 40000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 64, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "b55784db517142a58f06292ac7a077e4", "role": "primary", "sha256": "da8662eb1a56aa5ae6557f7d85b553d384f18a4c25bf7931602ab9d7bfe7e795", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "8x64000-cold", "cell_outcome_sha256": "9aac3194f2d288d61c5e86450fbcff35e993f47bd7eb3acc9740bb8ec7ff65ef", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "6af2e6e68e5a40adbe437f29", "licence_class": "commercial", "measured_at": "2026-10-18T22:10:41.479674+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 6.87569573185367 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 4.197763812740735 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 29.697188519750156 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 106.83074527746066 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 128.0724464179948 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 3.9061477494520407 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 3.9380458108476684 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 3.77913865315309 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 4.099310075263326 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 4.386106949436853 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 4.296217550218144 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 48.528760785121825 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 48.18503205232095 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 68.09 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.002490368708802264 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 295538.56720034726 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 4.259772005934014 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 29.195475804392792 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 106.81424680154305 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 127.41680157184601 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 3.6387653255067782 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 4.010408705018851 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 4.219142603190393 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 4.218973483888987 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 4.380743778799746 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 4.300401408677634 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 48.440925474889404 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 49.011069344675114 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 71.3 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0026152224289207895 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 284105.8534002441 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 6.87569573185367 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 42.21321325065051 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 116.00782173697371 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 128.0662279399112 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 5.935582325921318 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 6.991263832650078 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 6.7601276310572604 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 6.6852505672590326 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 5.774505875023833 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 7.104660548296282 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 7.661812951049524 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 49.80609180474219 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 70.41 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.002652863782727232 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 277812.9826335597 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[3].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 7.028005996415999 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[3].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 42.77518180605785 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[3].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 115.99130118894391 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[3].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 128.04524536104873 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[3].decode_overlap", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 5.537166416019333 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 6.870671530801338 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 6.228805918376998 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 7.084363849597519 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 6.995692296093376 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 7.060319696738623 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 7.858530214575379 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 49.90014504167757 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[3].power_w_median", "statistic": "median", "unit": "W", "value": 70.84 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[3].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0026753257848995565 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[3].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 279965.9033032946 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[4].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 7.069985982886312 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[4].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 42.07623129452966 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[4].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 115.99527246400248 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[4].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 128.08083567488939 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[4].decode_overlap", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 6.5300836825534585 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 6.128051547412834 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 5.5424309463870545 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 7.140075138669161 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 7.062616743962466 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 7.405820365188401 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 7.077355221810158 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 51.23315068730694 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[4].power_w_median", "statistic": "median", "unit": "W", "value": 71.75 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[4].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.002690617438115672 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[4].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 272428.17563590314 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 8, "configured_context_tokens": 262144, "context_tokens": 64000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 8, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": "observed-throughput-boundary", "classification": "capacity", "id": "6873366f56d14eadaccf5fc696aecbbd", "role": "primary", "sha256": "7d93186c6fe2cf486ca8e2d248c0fca064137f0280d73630293467f41cae0765", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, { "capacity_kind": "observed-throughput-boundary", "classification": "capacity", "id": "656408afe89d4bfaa54f338fb040aee1", "role": "confirmation", "sha256": "804b4cd3684668ae12890d6c8c09c755e344205da585720291303a836fb9b79f", "variant_sha256": "ce7a11417103cc42fa9c8145854c46461d126b984c768acf8e243600a467a158" } ], "cell_id": "16x64000-cold", "cell_outcome_sha256": "e3baafba6bdabbde2673d14ab0973413053860df8f06c4b488d6a28d55b36ec5", "cell_status": "capacity-limit-confirmed", "limit_reference": null, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "8041f2927d9a4192a985c7c3", "licence_class": "commercial", "measured_at": "2026-10-18T22:27:56.251745+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "cold", "concurrency": 16, "configured_context_tokens": 262144, "context_tokens": 64000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 16, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "57c6d088c6294285b4fd27d25072aad6", "role": "primary", "sha256": "655bfe5824750cb3f6c6092eb3f56917116308f1b943d711d73070d365192786", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "5x64000-cold", "cell_outcome_sha256": "09ac217aa50a83bd2ce245f3772b6d01236418ee630495d089eb01500f3a9971", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "86b713dcaddb4c7090a64bc0", "licence_class": "commercial", "measured_at": "2026-10-18T22:00:38.847366+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 44.57820668642665 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 44.16234533985952 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 204.13795284848996 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 78.23271867306903 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 78.23354443302378 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 45.15076650606327 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 44.077440141622766 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 43.82353487229463 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 44.71558773686318 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 44.16234533985952 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 65.21 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0015425460547964377 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 296911.7185032378 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 44.57820668642665 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 127.7420444256673 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 79.40107471495867 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 79.40232364600524 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 44.57820668642665 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 44.62313757289143 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 26.89529491082609 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 43.654699295351364 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 45.90470683153235 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 67.51 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0016180141788724265 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 283063.0324384267 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 44.67265876423206 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 111.80419631062891 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 77.90672580781393 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 77.9320079931058 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 23.013795377369267 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 44.55941768797434 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 45.11928714895216 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 44.67265876423206 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 45.70885095460589 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 69.55 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0016439581364510064 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 277987.6140803538 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 64000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 5, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "3a3a121dbf5f44f69f98de73bed7a358", "role": "primary", "sha256": "f38782e538d4ad135cd4cb3d687ebb142bc4867583d1d64e893e115dc9aad99c", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "8x4000-cold", "cell_outcome_sha256": "bb17a8770bc85adb3c44c168446e467e15bdb881e979f1dedbab3799bcc4efd4", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "88c14bb0493e49208785570b", "licence_class": "commercial", "measured_at": "2026-10-18T19:46:23.059075+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 43.17246593756465 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 40.86344994033777 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 224.23974751507686 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 5.527706830063835 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 6.0912731578573585 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 28.768517927396612 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 40.90209924023731 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 52.8977962239631 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 40.4258454244458 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 40.455416895633284 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 40.82480064043824 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 41.28931531406536 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 53.4585054537693 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 12.03 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0001592006615446198 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 4466061.843598088 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 43.17246593756465 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 261.80405637867705 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 4.384278599056415 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 4.781520818127319 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 43.01170594772752 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 35.11513038371296 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 42.93744016719711 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 43.17980609712562 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 43.16512577800369 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 43.51930562564278 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 51.9610430642149 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 55.11452856136612 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 12.1 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00015733834077371445 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 4607904.191914049 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 44.04111390232174 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 273.74455843005603 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 4.384028275962919 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 4.783547303872183 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 35.52933221897537 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 52.608176321957956 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 44.13485787500211 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 43.6694443671669 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 43.93731206810455 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 43.94736992964137 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 52.67591109036404 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 44.334982337614264 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 12.16 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0001575407851726153 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 4513117.02693983 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 8, "configured_context_tokens": 262144, "context_tokens": 4000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 8, "warmup": null } }, { "adaptive_cell": { "attempts": [], "cell_id": "32x40000-cold", "cell_outcome_sha256": "f6ed006eb42c989619b4e8f03c5e13e382d47172fdbd1a82000c2a70f66969bd", "cell_status": "above-confirmed-limit", "limit_reference": { "attempt_sha256": [ "aee01025e9c568df2b02c38b20918b42eb99e0da86fd5111099cb8d7c9c5cb93", "777aff85ef232ae148a855cb769c47394647360313ff9e999ad3c71e99fb8638" ], "cell_id": "16x40000-cold", "payload_sha256": "442b5f767aec4fb4d71a3c1945592a8efdbf5a7aa308a9f2e988d527b5eee9ae" }, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "8db184e1e1e642efb3da4030", "licence_class": "commercial", "measured_at": "2026-10-18T21:44:27.288340+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "cold", "concurrency": 32, "configured_context_tokens": 262144, "context_tokens": 40000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 32, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": "observed-serialization-boundary", "classification": "capacity", "id": "63231ed73a1145ca84a60c5d29912207", "role": "primary", "sha256": "51ec5c7fa01a6e5f9b4dc61cdad0bf943e848fa45a62b4e7e75b2e5f22b1ced8", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, { "capacity_kind": null, "classification": "measured", "id": "046e496c617241b1b7a31265ef0c9641", "role": "confirmation", "sha256": "650495fee713060896c55d5c526baa2784ac0ad0e2825075d74c850af8c5f2a1", "variant_sha256": "ce7a11417103cc42fa9c8145854c46461d126b984c768acf8e243600a467a158" } ], "cell_id": "32x4000-cold", "cell_outcome_sha256": "7b93a8ec94506499f1376e6f8aed76a4176ad98d392bd9ea385a4acfbb2ffd13", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "ce7a11417103cc42fa9c8145854c46461d126b984c768acf8e243600a467a158" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "93f3eb0f88f141a897062601", "licence_class": "commercial", "measured_at": "2026-10-18T20:06:17.782400+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 9.770231462270724 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 9.555478249388203 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 157.3984556663292 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 12.974610606906936 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 19.38623060612008 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 5.100715303149196 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 5.357817296777526 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 5.295956790461295 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 5.545419857194718 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 6.219681489115597 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 6.328708513981551 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 6.878364305315372 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 6.293214090598395 }, { "basis": null, "n": null, "name": "per_person_decode_tps[8]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[8]", "statistic": "single", "unit": "tok/s", "value": 6.837144076529183 }, { "basis": null, "n": null, "name": "per_person_decode_tps[9]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[9]", "statistic": "single", "unit": "tok/s", "value": 5.7094596000500495 }, { "basis": null, "n": null, "name": "per_person_decode_tps[10]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[10]", "statistic": "single", "unit": "tok/s", "value": 9.015420813228934 }, { "basis": null, "n": null, "name": "per_person_decode_tps[11]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[11]", "statistic": "single", "unit": "tok/s", "value": 7.104612310458669 }, { "basis": null, "n": null, "name": "per_person_decode_tps[12]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[12]", "statistic": "single", "unit": "tok/s", "value": 8.35637574578052 }, { "basis": null, "n": null, "name": "per_person_decode_tps[13]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[13]", "statistic": "single", "unit": "tok/s", "value": 9.377252728105733 }, { "basis": null, "n": null, "name": "per_person_decode_tps[14]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[14]", "statistic": "single", "unit": "tok/s", "value": 7.696650673495624 }, { "basis": null, "n": null, "name": "per_person_decode_tps[15]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[15]", "statistic": "single", "unit": "tok/s", "value": 5.817957241708693 }, { "basis": null, "n": null, "name": "per_person_decode_tps[16]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[16]", "statistic": "single", "unit": "tok/s", "value": 10.552991013461831 }, { "basis": null, "n": null, "name": "per_person_decode_tps[17]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[17]", "statistic": "single", "unit": "tok/s", "value": 10.722336051214565 }, { "basis": null, "n": null, "name": "per_person_decode_tps[18]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[18]", "statistic": "single", "unit": "tok/s", "value": 10.723018754144565 }, { "basis": null, "n": null, "name": "per_person_decode_tps[19]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[19]", "statistic": "single", "unit": "tok/s", "value": 9.733703770670674 }, { "basis": null, "n": null, "name": "per_person_decode_tps[20]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[20]", "statistic": "single", "unit": "tok/s", "value": 18.308623914210994 }, { "basis": null, "n": null, "name": "per_person_decode_tps[21]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[21]", "statistic": "single", "unit": "tok/s", "value": 21.030723961933482 }, { "basis": null, "n": null, "name": "per_person_decode_tps[22]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[22]", "statistic": "single", "unit": "tok/s", "value": 18.346432928075274 }, { "basis": null, "n": null, "name": "per_person_decode_tps[23]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[23]", "statistic": "single", "unit": "tok/s", "value": 21.504744326226625 }, { "basis": null, "n": null, "name": "per_person_decode_tps[24]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[24]", "statistic": "single", "unit": "tok/s", "value": 23.944013029659747 }, { "basis": null, "n": null, "name": "per_person_decode_tps[25]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[25]", "statistic": "single", "unit": "tok/s", "value": 21.539753398283764 }, { "basis": null, "n": null, "name": "per_person_decode_tps[26]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[26]", "statistic": "single", "unit": "tok/s", "value": 13.11395553922347 }, { "basis": null, "n": null, "name": "per_person_decode_tps[27]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[27]", "statistic": "single", "unit": "tok/s", "value": 26.016774112652126 }, { "basis": null, "n": null, "name": "per_person_decode_tps[28]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[28]", "statistic": "single", "unit": "tok/s", "value": 26.224947651347193 }, { "basis": null, "n": null, "name": "per_person_decode_tps[29]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[29]", "statistic": "single", "unit": "tok/s", "value": 16.15937245466759 }, { "basis": null, "n": null, "name": "per_person_decode_tps[30]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[30]", "statistic": "single", "unit": "tok/s", "value": 19.242760665504974 }, { "basis": null, "n": null, "name": "per_person_decode_tps[31]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[31]", "statistic": "single", "unit": "tok/s", "value": 27.37990217365741 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[8]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[8]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[9]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[9]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[10]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[10]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[11]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[11]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[12]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[12]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[13]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[13]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[14]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[14]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[15]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[15]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[16]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[16]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[17]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[17]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[18]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[18]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[19]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[19]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[20]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[20]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[21]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[21]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[22]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[22]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[23]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[23]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[24]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[24]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[25]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[25]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[26]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[26]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[27]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[27]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[28]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[28]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[29]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[29]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[30]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[30]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[31]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[31]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 12.305 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0004720481241644538 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 6177336.272994305 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 9.770231462270724 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 157.5906844897033 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 12.692551162093878 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 19.096720915054902 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 5.136581606186466 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 5.457323390956383 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 5.427248995788918 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 5.549554708313949 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 6.1519077601652095 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 5.425402512288499 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 5.584834577592822 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 6.889551425871171 }, { "basis": null, "n": null, "name": "per_person_decode_tps[8]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[8]", "statistic": "single", "unit": "tok/s", "value": 8.106737220344796 }, { "basis": null, "n": null, "name": "per_person_decode_tps[9]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[9]", "statistic": "single", "unit": "tok/s", "value": 6.963559390276915 }, { "basis": null, "n": null, "name": "per_person_decode_tps[10]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[10]", "statistic": "single", "unit": "tok/s", "value": 7.041878768057966 }, { "basis": null, "n": null, "name": "per_person_decode_tps[11]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[11]", "statistic": "single", "unit": "tok/s", "value": 9.386219028564865 }, { "basis": null, "n": null, "name": "per_person_decode_tps[12]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[12]", "statistic": "single", "unit": "tok/s", "value": 7.715309834666177 }, { "basis": null, "n": null, "name": "per_person_decode_tps[13]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[13]", "statistic": "single", "unit": "tok/s", "value": 10.6879166457718 }, { "basis": null, "n": null, "name": "per_person_decode_tps[14]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[14]", "statistic": "single", "unit": "tok/s", "value": 10.569795651036962 }, { "basis": null, "n": null, "name": "per_person_decode_tps[15]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[15]", "statistic": "single", "unit": "tok/s", "value": 9.427524060772088 }, { "basis": null, "n": null, "name": "per_person_decode_tps[16]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[16]", "statistic": "single", "unit": "tok/s", "value": 6.731231535373867 }, { "basis": null, "n": null, "name": "per_person_decode_tps[17]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[17]", "statistic": "single", "unit": "tok/s", "value": 6.84194224306603 }, { "basis": null, "n": null, "name": "per_person_decode_tps[18]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[18]", "statistic": "single", "unit": "tok/s", "value": 12.421884636167524 }, { "basis": null, "n": null, "name": "per_person_decode_tps[19]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[19]", "statistic": "single", "unit": "tok/s", "value": 15.095637016590537 }, { "basis": null, "n": null, "name": "per_person_decode_tps[20]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[20]", "statistic": "single", "unit": "tok/s", "value": 18.09097136794963 }, { "basis": null, "n": null, "name": "per_person_decode_tps[21]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[21]", "statistic": "single", "unit": "tok/s", "value": 21.088317395452005 }, { "basis": null, "n": null, "name": "per_person_decode_tps[22]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[22]", "statistic": "single", "unit": "tok/s", "value": 18.568436051035036 }, { "basis": null, "n": null, "name": "per_person_decode_tps[23]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[23]", "statistic": "single", "unit": "tok/s", "value": 25.630052638705152 }, { "basis": null, "n": null, "name": "per_person_decode_tps[24]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[24]", "statistic": "single", "unit": "tok/s", "value": 10.11293886376936 }, { "basis": null, "n": null, "name": "per_person_decode_tps[25]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[25]", "statistic": "single", "unit": "tok/s", "value": 21.093966770348455 }, { "basis": null, "n": null, "name": "per_person_decode_tps[26]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[26]", "statistic": "single", "unit": "tok/s", "value": 11.430175282477963 }, { "basis": null, "n": null, "name": "per_person_decode_tps[27]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[27]", "statistic": "single", "unit": "tok/s", "value": 23.648420352366287 }, { "basis": null, "n": null, "name": "per_person_decode_tps[28]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[28]", "statistic": "single", "unit": "tok/s", "value": 21.946015238467847 }, { "basis": null, "n": null, "name": "per_person_decode_tps[29]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[29]", "statistic": "single", "unit": "tok/s", "value": 18.89864230263017 }, { "basis": null, "n": null, "name": "per_person_decode_tps[30]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[30]", "statistic": "single", "unit": "tok/s", "value": 26.27152977624748 }, { "basis": null, "n": null, "name": "per_person_decode_tps[31]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[31]", "statistic": "single", "unit": "tok/s", "value": 27.480958111830628 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[8]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[8]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[9]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[9]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[10]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[10]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[11]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[11]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[12]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[12]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[13]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[13]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[14]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[14]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[15]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[15]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[16]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[16]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[17]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[17]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[18]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[18]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[19]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[19]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[20]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[20]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[21]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[21]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[22]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[22]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[23]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[23]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[24]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[24]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[25]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[25]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[26]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[26]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[27]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[27]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[28]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[28]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[29]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[29]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[30]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[30]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[31]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[31]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 12.44 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0004774858844719161 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 6104892.510537554 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 9.780940423489803 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 157.7142631386363 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 12.790273281978443 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 19.206124195130542 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 5.095251424821244 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 5.383897067171134 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 5.445932835631976 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 5.416514943435728 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 5.539443530072917 }, { "basis": null, "n": null, "name": "per_person_decode_tps[5]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[5]", "statistic": "single", "unit": "tok/s", "value": 5.539129433014484 }, { "basis": null, "n": null, "name": "per_person_decode_tps[6]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[6]", "statistic": "single", "unit": "tok/s", "value": 6.323768443550663 }, { "basis": null, "n": null, "name": "per_person_decode_tps[7]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[7]", "statistic": "single", "unit": "tok/s", "value": 6.873390236373518 }, { "basis": null, "n": null, "name": "per_person_decode_tps[8]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[8]", "statistic": "single", "unit": "tok/s", "value": 6.871849338185627 }, { "basis": null, "n": null, "name": "per_person_decode_tps[9]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[9]", "statistic": "single", "unit": "tok/s", "value": 6.393730646145335 }, { "basis": null, "n": null, "name": "per_person_decode_tps[10]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[10]", "statistic": "single", "unit": "tok/s", "value": 7.025675413370198 }, { "basis": null, "n": null, "name": "per_person_decode_tps[11]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[11]", "statistic": "single", "unit": "tok/s", "value": 10.492186535440746 }, { "basis": null, "n": null, "name": "per_person_decode_tps[12]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[12]", "statistic": "single", "unit": "tok/s", "value": 9.362595285282241 }, { "basis": null, "n": null, "name": "per_person_decode_tps[13]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[13]", "statistic": "single", "unit": "tok/s", "value": 6.855906029795853 }, { "basis": null, "n": null, "name": "per_person_decode_tps[14]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[14]", "statistic": "single", "unit": "tok/s", "value": 10.645903252922384 }, { "basis": null, "n": null, "name": "per_person_decode_tps[15]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[15]", "statistic": "single", "unit": "tok/s", "value": 8.520571717822456 }, { "basis": null, "n": null, "name": "per_person_decode_tps[16]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[16]", "statistic": "single", "unit": "tok/s", "value": 7.943735894042069 }, { "basis": null, "n": null, "name": "per_person_decode_tps[17]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[17]", "statistic": "single", "unit": "tok/s", "value": 9.703004792228187 }, { "basis": null, "n": null, "name": "per_person_decode_tps[18]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[18]", "statistic": "single", "unit": "tok/s", "value": 17.326203962807277 }, { "basis": null, "n": null, "name": "per_person_decode_tps[19]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[19]", "statistic": "single", "unit": "tok/s", "value": 9.858876054751416 }, { "basis": null, "n": null, "name": "per_person_decode_tps[20]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[20]", "statistic": "single", "unit": "tok/s", "value": 17.972667434026803 }, { "basis": null, "n": null, "name": "per_person_decode_tps[21]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[21]", "statistic": "single", "unit": "tok/s", "value": 21.11592738024244 }, { "basis": null, "n": null, "name": "per_person_decode_tps[22]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[22]", "statistic": "single", "unit": "tok/s", "value": 23.951163408729162 }, { "basis": null, "n": null, "name": "per_person_decode_tps[23]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[23]", "statistic": "single", "unit": "tok/s", "value": 25.386955741527558 }, { "basis": null, "n": null, "name": "per_person_decode_tps[24]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[24]", "statistic": "single", "unit": "tok/s", "value": 18.254860156283307 }, { "basis": null, "n": null, "name": "per_person_decode_tps[25]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[25]", "statistic": "single", "unit": "tok/s", "value": 13.051027466341118 }, { "basis": null, "n": null, "name": "per_person_decode_tps[26]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[26]", "statistic": "single", "unit": "tok/s", "value": 21.380651777889486 }, { "basis": null, "n": null, "name": "per_person_decode_tps[27]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[27]", "statistic": "single", "unit": "tok/s", "value": 11.43658381516285 }, { "basis": null, "n": null, "name": "per_person_decode_tps[28]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[28]", "statistic": "single", "unit": "tok/s", "value": 25.7157367388493 }, { "basis": null, "n": null, "name": "per_person_decode_tps[29]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[29]", "statistic": "single", "unit": "tok/s", "value": 15.745639618288237 }, { "basis": null, "n": null, "name": "per_person_decode_tps[30]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[30]", "statistic": "single", "unit": "tok/s", "value": 21.763693626504963 }, { "basis": null, "n": null, "name": "per_person_decode_tps[31]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[31]", "statistic": "single", "unit": "tok/s", "value": 27.958285482778656 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[5]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[5]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[6]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[6]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[7]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[7]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[8]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[8]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[9]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[9]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[10]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[10]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[11]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[11]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[12]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[12]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[13]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[13]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[14]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[14]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[15]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[15]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[16]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[16]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[17]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[17]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[18]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[18]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[19]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[19]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[20]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[20]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[21]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[21]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[22]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[22]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[23]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[23]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[24]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[24]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[25]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[25]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[26]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[26]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[27]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[27]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[28]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[28]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[29]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[29]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[30]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[30]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[31]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[31]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 12.49 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0004895934549630578 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 5990684.659420762 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "ce7a11417103cc42fa9c8145854c46461d126b984c768acf8e243600a467a158", "cache_state": "cold", "concurrency": 32, "configured_context_tokens": 262144, "context_tokens": 4000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 32, "warmup": null } }, { "adaptive_cell": { "attempts": [], "cell_id": "32x64000-cold", "cell_outcome_sha256": "151e799025d436bc8062bf5ac438f487954709efee2c8c59d2d322bc5c5b5fa9", "cell_status": "above-confirmed-limit", "limit_reference": { "attempt_sha256": [ "7d93186c6fe2cf486ca8e2d248c0fca064137f0280d73630293467f41cae0765", "804b4cd3684668ae12890d6c8c09c755e344205da585720291303a836fb9b79f" ], "cell_id": "16x64000-cold", "payload_sha256": "e3baafba6bdabbde2673d14ab0973413053860df8f06c4b488d6a28d55b36ec5" }, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "a7c17cbaae874c6d93640050", "licence_class": "commercial", "measured_at": "2026-10-18T23:28:13.269320+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "cold", "concurrency": 32, "configured_context_tokens": 262144, "context_tokens": 64000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 32, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "b8f007eb9a5a4dd2a9ce9e78d05bd490", "role": "primary", "sha256": "a70d712b951ff9629acdcd5918af4e5dbb520be077f14d5f8f673c0b84f1c3a3", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "2x40000-cold", "cell_outcome_sha256": "eefec76c7f356086431ad0512f4c9f7f46b80c1e1d7cbd4d1d5ad002c6d8f2e7", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "ad1e722cebed41798c6e4de7", "licence_class": "commercial", "measured_at": "2026-10-18T20:52:39.887011+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 73.59540596746965 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 76.93389328039748 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 148.77376527794232 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 18.93542544543743 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 18.941689976956695 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 77.59286775848155 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 76.27491880231341 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 12.46 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00037782204529773024 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 526702.9874955671 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 73.59540596746965 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 137.70033951439765 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 17.39713887043763 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 17.440696423873305 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 71.28768441042064 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 75.90312752451867 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 12.74 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.000379036981186834 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 501270.34941307124 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 70.75077688439197 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 135.68814871077794 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 17.39191487140488 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 17.436043496010825 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 71.49219515082332 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 70.00935861796061 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 13.025 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00038573465970812313 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 492566.57450426876 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 2, "configured_context_tokens": 262144, "context_tokens": 40000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 2, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "b5096337976a4ccdb0119f1f2f74a2dc", "role": "primary", "sha256": "380ce4169ff88c1bfcc987fd6668c111c5e2207711aaf98d70f5c47524bb1695", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "5x128000-cold", "cell_outcome_sha256": "82802a0b932a2895e797d7b0b4974f9011033e4e8cb158aa9195f5e2dd10186b", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "afa0c78822e542038d9fb92e", "licence_class": "commercial", "measured_at": "2026-10-18T23:53:46.317817+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 25.019077723626758 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 36.00275450638482 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 158.44044993157797 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 223.7292455751449 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 223.73009509709664 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 33.6757365498962 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 36.31980387130431 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 35.89160890288309 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 36.00275450638482 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 37.98328007010752 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 76.95 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0047776828241582045 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 94606.53137426286 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 30.487335588151357 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 69.17610459591094 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 224.74697231710888 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 224.74851158307865 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 14.163422649851611 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 30.77913208196441 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 30.809371264017173 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 30.487335588151357 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 30.274423880688325 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 78.86 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.004999605720840404 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 89607.0660397384 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 25.019077723626758 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 47.59439419133189 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 222.36926944809966 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 223.10198389203288 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 9.617954189725223 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 17.972385902864623 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 25.019077723626758 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 32.015304645428394 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 32.73676871595134 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 80.38 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.005032169262368856 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 91610.59097264698 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[3].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 24.881234862591178 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[3].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 46.71803012863278 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[3].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 222.32929247897118 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[3].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 223.06492496305145 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[3].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 9.674062220690558 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 24.881234862591178 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 17.990997088060666 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 31.22741780990464 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 31.565336874371525 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[3].power_w_median", "statistic": "median", "unit": "W", "value": 80.23 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[3].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.005020649958086551 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[3].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 88235.5877621937 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[4].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 21.897429126539137 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[4].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 42.42814202004405 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[4].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 220.4858876860235 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[4].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 222.121685591992 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[4].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 8.688899079283138 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 14.848047468615096 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 32.78152848244248 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 33.345235902344754 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 21.897429126539137 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[4].power_w_median", "statistic": "median", "unit": "W", "value": 80.46 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[4].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.005005431900267337 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[4].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 90701.46373897369 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 128000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 5, "warmup": null } }, { "adaptive_cell": {}, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "b39ef8e95ec84d1d8edf318b", "licence_class": "commercial", "measured_at": "2026-10-18T19:29:18.960908+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "configured_context_tokens": 262144 }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "fad8fbcb0a87488ab124edc428be442e", "role": "primary", "sha256": "090f2c068c5f2fd531c52400cec40c85d83da11f329cb92cb48c2c8b77f15804", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "2x64000-cold", "cell_outcome_sha256": "7de148d7485adb8c603c8caa97189f2acb356d52b625a1e817670502e04f3163", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "b691ece4496e4f26bf9fa8ef", "licence_class": "commercial", "measured_at": "2026-10-18T21:51:28.418467+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 59.935247273144654 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 69.28591413051146 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 130.84551528392592 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 34.25204017898068 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 34.267043753992766 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 70.52297485050751 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 68.0488534105154 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 56.19 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0006786719009744738 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 263750.4215851314 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 59.38682590592751 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 98.65553873774267 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 33.60497915465385 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 33.82158480514772 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 52.635301471015346 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 66.13835034083968 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 57.79 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0007021171629806479 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 254943.2052623583 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 59.1176805224723 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 98.30681174143751 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 33.15501801203936 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 33.37191870203242 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 52.37798692067391 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 65.85737412427068 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 60.595 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.000710063562131567 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 252090.10790900612 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[3].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 59.935247273144654 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[3].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 101.98761380831914 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[3].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 33.20796163904015 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[3].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 33.4294022009708 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[3].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 52.57143319769552 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 67.29906134859378 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[3].power_w_median", "statistic": "median", "unit": "W", "value": 61.935 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[3].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0007147256984704029 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[3].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 250445.73097494754 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[4].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 60.47979552696644 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[4].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 101.4142288301267 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[4].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 33.23080122156534 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[4].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 33.44504476012662 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[4].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 52.60766833103974 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 68.35192272289315 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[4].power_w_median", "statistic": "median", "unit": "W", "value": 62.03 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[4].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.000721424841094615 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[4].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 248120.09485063478 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 2, "configured_context_tokens": 262144, "context_tokens": 64000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 2, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "583b5f03eb6c4cf18edd1439506d4d3c", "role": "primary", "sha256": "135193faedd09ee25647aa790459d27ee0fc2dde9fab493496bbfc1666bb7c13", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "1x40000-cold", "cell_outcome_sha256": "4b3ac346838862e804eaaaf9af91055c877d85a37ea3dc149e9368a7fb9555cb", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "c0bb23341b7e4f979f2cbcd4", "licence_class": "commercial", "measured_at": "2026-10-18T20:46:05.754752+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 93.70587620885446 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 94.5239741050509 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 94.52363166006036 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 11.395271144807339 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 11.395271144807339 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 94.5239741050509 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 12.355 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00024485892112868606 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 396146.4812181439 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 93.70587620885446 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 93.7055012700674 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 10.103360007982701 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 10.103360007982701 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 93.70587620885446 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 12.4 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00024258749613299637 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 358633.48848080385 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 91.83418168276886 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 91.83386059746161 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 10.100876952055842 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 10.100876952055842 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 91.83418168276886 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 12.56 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00024529676356576903 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 395439.3795904785 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 40000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "4063dfed16eb4593a637684405b6a5c2", "role": "primary", "sha256": "4de56b089fc34a9450e66070a2827c30bd856a511cefc85b263f8ad5bfe5c4f1", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "1x4000-cold", "cell_outcome_sha256": "5f9b1f3eb7d835c459995d73269b5ffebb13990b08c1c4a573326172a89cff3c", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "c19a5dc4881040609806a8d3", "licence_class": "commercial", "measured_at": "2026-10-18T19:29:18.969968+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 104.04195006111404 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 104.92396389483945 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 104.92372380510635 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 2.1158086308278143 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 2.1158086308278143 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 104.92396389483945 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 12.05 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 9.574405161509154e-05 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 1013117.7693415106 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 104.04195006111404 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 104.04156081451708 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 0.773807352874428 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 0.773807352874428 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 104.04195006111404 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 12.05 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 9.131542261306831e-05 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 1171762.6326211302 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 103.55658109265264 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 103.5560834120746 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 0.7619801149703562 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 0.7619801149703562 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 103.55658109265264 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 12.06 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 9.080581829375846e-05 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 1068213.4892084035 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 4000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "adaptive_cell": { "attempts": [], "cell_id": "20x40000-cold", "cell_outcome_sha256": "f318470bd3577d4b486855c3cbc424cfc87485097ef9d2ad599652b515688d6f", "cell_status": "above-confirmed-limit", "limit_reference": { "attempt_sha256": [ "aee01025e9c568df2b02c38b20918b42eb99e0da86fd5111099cb8d7c9c5cb93", "777aff85ef232ae148a855cb769c47394647360313ff9e999ad3c71e99fb8638" ], "cell_id": "16x40000-cold", "payload_sha256": "442b5f767aec4fb4d71a3c1945592a8efdbf5a7aa308a9f2e988d527b5eee9ae" }, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "c6eb397419a5418c85b9b1cc", "licence_class": "commercial", "measured_at": "2026-10-18T21:44:27.129155+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "cold", "concurrency": 20, "configured_context_tokens": 262144, "context_tokens": 40000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 20, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "5735f4e45a424862a10084eaacfa9558", "role": "primary", "sha256": "2a491ddd5dff830b6b65124b009e695c3b48a7c021047dc55cbd49cbadca6e68", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "5x4000-cold", "cell_outcome_sha256": "1bee83935d1b1f813d2175ae9a3fba427139230ec20f5ebfe1c454c27eb8b611", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "c805d6d0f0324750a2fbff21", "licence_class": "commercial", "measured_at": "2026-10-18T19:40:34.295333+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 65.78076431267863 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 65.67935979236972 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 320.50819321764743 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 3.9385539640206844 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 3.9391679151449353 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 65.74711010350484 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 65.72122502786746 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 65.6510115734007 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 65.67935979236972 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 65.47588995317301 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 12.19 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00013269226196597122 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 3413914.219927695 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 65.78076431267863 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 285.27748106408245 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 2.936815827852115 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 2.937850706046447 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 65.85432792860911 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 65.83578636919863 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 61.2343254374907 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 65.78076431267863 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 65.30839554777097 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 12.22 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00012970599862931122 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 3569611.3124514376 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 65.93245558766847 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 291.70062116875454 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 2.9419198469258845 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 2.942558502079919 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 65.92319551157209 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 65.95807897917167 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 65.93245558766847 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 61.110502550452395 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 67.16797398101998 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 12.25 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0001296890983175315 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 3570076.4829622633 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 4000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 5, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "4aa9ae98745f4c8e8d5eba558f292350", "role": "primary", "sha256": "f08826e4343f72b6d8a5fad89d023ebe15d2cb32b12259c9fe767546f0410d1e", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "2x4000-cold", "cell_outcome_sha256": "7fb649c044ab39dc2b14b1391bf67cc7dd52555265dcf22b401a5a9377c9acc5", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "e4af62d7a53c4c129c02fb5f", "licence_class": "commercial", "measured_at": "2026-10-18T19:34:55.890352+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 86.44179276731317 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 83.19539961483467 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 151.25023881030367 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 2.8485600233543664 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 2.8508466898929328 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 81.8816143612499 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 84.50918486841944 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 12.2 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00010942201479205461 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 1791230.0406136557 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 86.44179276731317 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 160.93499756351176 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 1.3359154395293444 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 1.35632364009507 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 85.64709081816066 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 87.23649471646569 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 12.23 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00010151368369587387 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 1792861.7440901475 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 86.84273839073333 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 161.87994466175263 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 1.33603469398804 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 1.3598098871298134 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 88.72081106829631 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 84.96466571317036 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 12.16 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00010107097359411743 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 1810608.8572461426 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 2, "configured_context_tokens": 262144, "context_tokens": 4000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 2, "warmup": null } }, { "adaptive_cell": { "attempts": [], "cell_id": "64x64000-cold", "cell_outcome_sha256": "1a471676b4fbacbaafe88b887a5681a6a8bf07becea140baceda66b0ef3e0895", "cell_status": "above-confirmed-limit", "limit_reference": { "attempt_sha256": [ "7d93186c6fe2cf486ca8e2d248c0fca064137f0280d73630293467f41cae0765", "804b4cd3684668ae12890d6c8c09c755e344205da585720291303a836fb9b79f" ], "cell_id": "16x64000-cold", "payload_sha256": "e3baafba6bdabbde2673d14ab0973413053860df8f06c4b488d6a28d55b36ec5" }, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "e69269eed50c42e49abfcb61", "licence_class": "commercial", "measured_at": "2026-10-18T23:28:13.471750+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "cold", "concurrency": 64, "configured_context_tokens": 262144, "context_tokens": 64000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 64, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "9aaf477bb80d437996b48f09ad5aa26a", "role": "primary", "sha256": "9ee4fa00e88e14f34877da28a22c09543f0bc6ee4e116529b1e80ee7c984a79d", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "1x64000-cold", "cell_outcome_sha256": "ceec48b201856f401842a9997059c702ea173a1a89ecf8fe314cd0392b38fa26", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "f059dc093383456680cc9f50", "licence_class": "commercial", "measured_at": "2026-10-18T21:44:27.612508+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 88.60880361250577 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 88.81617975822793 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 88.81607909855624 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 19.881602228153497 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 19.881602228153497 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 88.81617975822793 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 13.01 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0004039098665841991 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 232725.18890153122 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 88.45995402673495 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 88.4596842169869 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 18.892973887966946 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 18.892973887966946 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 88.45995402673495 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 13.325 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00040938788769847694 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 219840.40736028555 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 88.60880361250577 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 88.60857704264279 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 18.913371693110093 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 18.913371693110093 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 88.60880361250577 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 13.5 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00041791745485697307 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 239281.7022543931 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-18", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 64000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "adaptive_cell": {}, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "10af912e6e8b40a29d44cf9a", "licence_class": "commercial", "measured_at": "2026-10-19T07:37:31.789511+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "configured_context_tokens": 262144 }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "adaptive_cell": {}, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "190a389760114554bbe10552", "licence_class": "commercial", "measured_at": "2026-10-19T09:52:50.059518+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "pages_per_hour", "scope": "machine_total", "source_field": "standard_result.pages_per_hour", "statistic": "single", "unit": "count", "value": 8906.194696845945 }, { "basis": null, "n": null, "name": "bytes_per_1000_pages", "scope": "machine_total", "source_field": "standard_result.bytes_per_1000_pages", "statistic": "single", "unit": "count", "value": 312316695.0 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "configured_context_tokens": 262144 }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "adaptive_cell": {}, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "192f0255ca564e619ee16f0d", "licence_class": "commercial", "measured_at": "2026-10-19T08:20:04.794872+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "configured_context_tokens": 262144 }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "adaptive_cell": { "attempts": [], "cell_id": "64x250000-warm", "cell_outcome_sha256": "69504fb452827e2906d6782f3b63e7f272ecf8f07a42d0493af79ef5f8d8b08a", "cell_status": "above-confirmed-limit", "limit_reference": { "attempt_sha256": [ "291e5591b60d97dba4f6913080ce7fab17ad1937b73ac471d6f425205862f2fe", "eb26c7f688d2de53cc76ca52e268e0960055e31e2b8db92a2dabc6da21d75aa1" ], "cell_id": "8x250000-warm", "payload_sha256": "059e2470391458d71c724eaed8cd41b0f975e3544f97606d8f5550381bf372b1" }, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "1c5a757c520c476abc2ef20e", "licence_class": "commercial", "measured_at": "2026-10-19T07:37:31.306864+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "warm", "concurrency": 64, "configured_context_tokens": 262144, "context_tokens": 250000, "scenario": "shared" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": null, "document": "shared", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 64, "warmup": null } }, { "caveats": [ "Operator-supplied runtime manifest; the endpoint runner cannot attest hardware or model pins. Bounded task correctness is not customer acceptance.", "Engine version was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": null }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "2161604163794ac4afe4dea4", "licence_class": "commercial", "measured_at": "2026-10-19T08:00:08.965685+00:00", "metrics": [ { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 32 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 26 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[0].concurrency", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[0].passed", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[0].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "concurrency", "scope": "machine_total", "source_field": "workload_levels[1].concurrency", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "workload_levels[1].passed", "statistic": "single", "unit": "count", "value": 13 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "workload_levels[1].total", "statistic": "single", "unit": "count", "value": 16 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "unrecorded" }, "outcome": { "gates": [ { "met": false, "name": "structured-tasks", "observed": "26/32" } ], "reasons": [ "26/32 tasks passed." ], "status": "partial" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 26, "skipped": null, "suite": "structured-tasks", "total": 32, "version": null } ], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "workload-qualification", "version": null }, "settings": { "context_tokens": 262144, "kv_dtype": "fp8", "speculative_decoding": "with multi-token prediction" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "adaptive_cell": { "attempts": [], "cell_id": "32x250000-cold", "cell_outcome_sha256": "e30eaf369608060062fd4cef78f69c213fb21f0df6c5cd2e33163c1efa256275", "cell_status": "above-confirmed-limit", "limit_reference": { "attempt_sha256": [ "e9093e7223960cdffc52b7bb5e2364f5161032d61ed97d7136415d29de1b6bff", "8be552f9423936688b93831ed32f0e11f70f0ba807f2e1d37dc96d6252959f60" ], "cell_id": "8x250000-cold", "payload_sha256": "24c64ed72bb2a1c28206388bf5c8d08a76ff0ee506cfb3191efeaf4053ebe7df" }, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "21e0a1847eb341afbd52f197", "licence_class": "commercial", "measured_at": "2026-10-19T04:57:04.285980+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "cold", "concurrency": 32, "configured_context_tokens": 262144, "context_tokens": 250000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 32, "warmup": null } }, { "adaptive_cell": {}, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "24192130b83d6c60af6e5886", "licence_class": "commercial", "measured_at": "2026-10-19T09:54:04.706498+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "configured_context_tokens": 262144 }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "20497980849b4f5a83b47909452b73e8", "role": "primary", "sha256": "39aa4313b0e81d4b26d1315b3a1809a64d01fee9de093f7dea0f8e2b141197eb", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "1x250000-cold", "cell_outcome_sha256": "e5d8dbc96c8e4cc5d2ad1e11b30738421f80a25c0f5b0898d8c0045f77ce09fa", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "2d74ac02f206445faaa2f343", "licence_class": "commercial", "measured_at": "2026-10-19T01:23:30.145913+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 61.96243656915724 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 60.99471305028971 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 60.99456240099517 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 156.77964339009486 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 156.77964339009486 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 60.99471305028971 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 73.39 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0032505767258376107 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 26456.843586068397 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 62.42650171869672 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 62.42639554846983 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 156.26623678207397 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 156.26623678207397 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 62.42650171869672 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 76.48 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0033917356191286085 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 28009.258582603507 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 61.96243656915724 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 61.962320962825736 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 156.46255602198653 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 156.46255602198653 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 61.96243656915724 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 77.15 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0034352393616851272 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 26199.04772977778 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 250000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "adaptive_cell": { "attempts": [], "cell_id": "64x250000-cold", "cell_outcome_sha256": "ed1152698872809a2893080b1cf624f7cdc026555db580406da834fc108fe792", "cell_status": "above-confirmed-limit", "limit_reference": { "attempt_sha256": [ "e9093e7223960cdffc52b7bb5e2364f5161032d61ed97d7136415d29de1b6bff", "8be552f9423936688b93831ed32f0e11f70f0ba807f2e1d37dc96d6252959f60" ], "cell_id": "8x250000-cold", "payload_sha256": "24c64ed72bb2a1c28206388bf5c8d08a76ff0ee506cfb3191efeaf4053ebe7df" }, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "2f1a99e4887440b881653034", "licence_class": "commercial", "measured_at": "2026-10-19T04:57:04.558120+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "cold", "concurrency": 64, "configured_context_tokens": 262144, "context_tokens": 250000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 64, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": "observed-throughput-boundary", "classification": "capacity", "id": "8c733ec0c7704cebbae5bf332549bf12", "role": "primary", "sha256": "6579ba6d1cc9de0e9fb868caa8e2c9f38d809d86f6c357b89478148b3fb4fc4a", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, { "capacity_kind": "observed-throughput-boundary", "classification": "capacity", "id": "fe23f28dfc0646459927b440e7b6a6d0", "role": "confirmation", "sha256": "0ca8f125437dceb4b799ebcfc14a81188debe864b04ced14fceafa36960b3fa3", "variant_sha256": "ce7a11417103cc42fa9c8145854c46461d126b984c768acf8e243600a467a158" } ], "cell_id": "8x128000-cold", "cell_outcome_sha256": "cd9c683c823b57d6165353675fa9eb5f77c0fe6b29b52b61ce6bbed361100a74", "cell_status": "capacity-limit-confirmed", "limit_reference": null, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "328cc8c8b34b4653b662272b", "licence_class": "commercial", "measured_at": "2026-10-19T00:18:59.837819+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "cold", "concurrency": 8, "configured_context_tokens": 262144, "context_tokens": 128000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 8, "warmup": null } }, { "adaptive_cell": { "attempts": [], "cell_id": "20x128000-cold", "cell_outcome_sha256": "aa3f3f54344cf11e4d486950f9c14d96b13163ed9d2f1b37a9f81655361e24fc", "cell_status": "above-confirmed-limit", "limit_reference": { "attempt_sha256": [ "6579ba6d1cc9de0e9fb868caa8e2c9f38d809d86f6c357b89478148b3fb4fc4a", "0ca8f125437dceb4b799ebcfc14a81188debe864b04ced14fceafa36960b3fa3" ], "cell_id": "8x128000-cold", "payload_sha256": "cd9c683c823b57d6165353675fa9eb5f77c0fe6b29b52b61ce6bbed361100a74" }, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "3da91dd7b3364882afabd611", "licence_class": "commercial", "measured_at": "2026-10-19T01:23:29.386093+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "cold", "concurrency": 20, "configured_context_tokens": 262144, "context_tokens": 128000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 20, "warmup": null } }, { "adaptive_cell": {}, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "473e4da5db7d4875ad0ac997", "licence_class": "commercial", "measured_at": "2026-10-19T08:58:24.197030+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "configured_context_tokens": 262144 }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "05dd29437a074a3ab076db10897d183f", "role": "primary", "sha256": "a42c9f4ebff0da9d826b2123e01bdbeaf0c7c498690855d212f3bcdd76c39c83", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "2x250000-warm", "cell_outcome_sha256": "0d027b3a801643bc476cc4b39bcff1207cae7ca57deec3273d7f74ae721ac4d5", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "4b96f8f48c554cadb1226ef0", "licence_class": "commercial", "measured_at": "2026-10-19T05:11:58.481980+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 49.97597266113256 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 50.79107332023338 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 89.01721322207977 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 6.042079741950147 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 6.262960945954546 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 54.81035610277166 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 46.77179053769511 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 15.155000000000001 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0002510188919202515 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 948135.8083423155 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 49.89301629777245 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 85.15629948818207 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 6.026877856464125 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 6.236203354084864 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 54.86410025067967 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 44.921932344865226 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 15.07 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0002483807667143177 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 885736.8584139984 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 49.97597266113256 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 85.99375150960248 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 6.004167008562945 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 6.221503562992439 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 45.498947775853246 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 54.45299754641188 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 15.09 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.000246424897832082 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 896824.96348479 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "warm", "concurrency": 2, "configured_context_tokens": 262144, "context_tokens": 250000, "scenario": "shared" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": null, "document": "shared", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 2, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "0f7e50fe9cc447989a249da6ad04e422", "role": "primary", "sha256": "cbe005a37e323945b01f421c6465bcd689f9f29d19d1de1bc9bd735c67c2d905", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "5x250000-cold", "cell_outcome_sha256": "b04f3417700d0993c50b839cb0e0602cfb0bc9479edbabe6712647e48cca3ed5", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "51fd5c09268b4edcbbade598", "licence_class": "commercial", "measured_at": "2026-10-19T02:07:34.038430+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 18.81970168459099 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 18.81970168459099 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 61.438736729271405 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 679.3594120570924 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 680.1443467049394 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 17.77310472969827 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 21.416154139317857 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 21.505042905312905 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 13.04835440516925 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 18.81970168459099 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 84.07499999999999 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.015608356032255102 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 29471.393338888298 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 17.950314828173315 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 37.04097873777727 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 682.9090624400415 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 683.9191965109203 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 19.532400958123677 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 7.649437530367051 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 17.7799980140757 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 21.41384804735137 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 17.950314828173315 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 83.92 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.01583160202544532 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 28487.325494610774 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 19.683519138798292 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 26.465636052629186 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 680.952630587155 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 681.351193837123 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 19.683519138798292 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 5.638234277072666 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 14.351252740752747 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 21.059355901348304 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 19.888461442199965 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 83.85 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.015776414186824773 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 28333.434626310675 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 5, "warmup": null } }, { "caveats": [ "Bounded fictional tasks only. Unsupported cases are separately counted, never passes. Original publisher name with variant; exact variant repository and revision retained privately. Not customer acceptance.", "Publisher revision was not recorded for public reproduction.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "525c0716346b4c9b81526ec3", "licence_class": "commercial", "measured_at": "2026-10-19T08:01:15.141355+00:00", "metrics": [ { "basis": null, "n": null, "name": "skipped_cases", "scope": "machine_total", "source_field": "skipped_cases", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "correctness_failures", "scope": "machine_total", "source_field": "correctness_failures", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "ttft_median_s", "scope": "per_person", "source_field": "ttft_median_s", "statistic": "median", "unit": "s", "value": 0.06368294009007514 }, { "basis": null, "n": null, "name": "workload_total", "scope": "machine_total", "source_field": "workload_total", "statistic": "single", "unit": "count", "value": 114 }, { "basis": null, "n": null, "name": "workload_passed", "scope": "machine_total", "source_field": "workload_passed", "statistic": "single", "unit": "count", "value": 111 }, { "basis": null, "n": null, "name": "api_cases_passed", "scope": "machine_total", "source_field": "api_cases_passed", "statistic": "single", "unit": "count", "value": 109 }, { "basis": null, "n": null, "name": "ttft_p95_s", "scope": "per_person", "source_field": "ttft_p95_s", "statistic": "p95", "unit": "s", "value": 0.09221734502352774 }, { "basis": null, "n": null, "name": "completion_p95_s", "scope": "per_person", "source_field": "completion_p95_s", "statistic": "p95", "unit": "s", "value": 0.5160631039179862 }, { "basis": null, "n": null, "name": "api_cases_total", "scope": "machine_total", "source_field": "api_cases_total", "statistic": "single", "unit": "count", "value": 112 }, { "basis": null, "n": null, "name": "completion_median_s", "scope": "per_person", "source_field": "completion_median_s", "statistic": "median", "unit": "s", "value": 0.1854163371026516 }, { "basis": null, "n": null, "name": "completeness_failures", "scope": "machine_total", "source_field": "completeness_failures", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[0].passed", "statistic": "single", "unit": "count", "value": 7 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[0].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[0].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[1].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[1].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[1].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[2].passed", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[2].total", "statistic": "single", "unit": "count", "value": 12 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[2].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[3].passed", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[3].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[3].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[4].passed", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[4].total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[4].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[5].passed", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[5].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[5].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[6].passed", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[6].total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[6].skipped", "statistic": "single", "unit": "count", "value": 4 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[7].passed", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[7].total", "statistic": "single", "unit": "count", "value": 6 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[7].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[8].passed", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[8].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[8].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[9].passed", "statistic": "single", "unit": "count", "value": 15 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[9].total", "statistic": "single", "unit": "count", "value": 16 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[9].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[10].passed", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[10].total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[10].skipped", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[11].passed", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[11].total", "statistic": "single", "unit": "count", "value": 8 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[11].skipped", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "passed", "scope": "machine_total", "source_field": "api_categories[12].passed", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "total", "scope": "machine_total", "source_field": "api_categories[12].total", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "skipped", "scope": "machine_total", "source_field": "api_categories[12].skipped", "statistic": "single", "unit": "count", "value": 4 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": false, "name": "structured-tasks", "observed": "111/114" }, { "met": false, "name": "API-cases", "observed": "109/112" } ], "reasons": [ "111/114 tasks passed.", "109/112 API cases passed." ], "status": "partial" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 111, "skipped": null, "suite": "structured-tasks", "total": 114, "version": "core-v2" }, { "passed": 109, "skipped": 16, "suite": "API-cases", "total": 112, "version": "core-v2" } ], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": "0ad0b13a94ad74bcdefe99d12d796b175f56e5777d9642a1dd4c0f170d708398", "harness": { "name": "keep-api-suite-run-v1", "version": "1" }, "harness_sha256": null, "os": null, "workload_name": "core-v2", "workload_version": "2" }, "requests": [], "scenario": { "id": "api-suite", "version": "1" }, "settings": { "concurrency": 1, "context_tokens": 262144, "kv_dtype": "fp8", "speculative_decoding": "with multi-token prediction" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "4bb84e4c29454e7da8621419a87fd87b", "role": "primary", "sha256": "d610a10a643c50873d34cd94a2ee596f52c9c529da7b20f20a1ef055d010ca5c", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "5x250000-warm", "cell_outcome_sha256": "67fb08f3cc8452d180e7522a0531d5360e7c658316890f1c88bc9aed1a0c66e5", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "568f7ea8aa5d4c56ae4103ec", "licence_class": "commercial", "measured_at": "2026-10-19T05:33:48.830719+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 35.942356025086376 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 35.61780867715461 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 78.94651772678276 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 14.693071058019996 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 14.694660450099036 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 35.61780867715461 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 16.32683808856185 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 35.72770463396678 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 35.79316116784128 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 27.363869568144302 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 14.975 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00045578199104487945 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 1167224.7049085703 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 35.942356025086376 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 79.16425181399548 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 14.416779736988246 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 14.417290423996747 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 25.210055856941743 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 16.425896948789028 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 36.33536821161394 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 36.37271974389951 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 35.942356025086376 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 14.98 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00045099392131556947 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 1148574.2390694995 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 36.22577084330709 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 80.15760875252553 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 14.45871768007055 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 14.459370015887544 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 25.355263790669103 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 16.448971494610863 }, { "basis": null, "n": null, "name": "per_person_decode_tps[2]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[2]", "statistic": "single", "unit": "tok/s", "value": 36.22577084330709 }, { "basis": null, "n": null, "name": "per_person_decode_tps[3]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[3]", "statistic": "single", "unit": "tok/s", "value": 36.748614363907166 }, { "basis": null, "n": null, "name": "per_person_decode_tps[4]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[4]", "statistic": "single", "unit": "tok/s", "value": 36.72853867544283 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[2]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[2]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[3]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[3]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[4]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[4]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 14.934999999999999 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0004515330829070923 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 1151632.1166371666 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "warm", "concurrency": 5, "configured_context_tokens": 262144, "context_tokens": 250000, "scenario": "shared" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": null, "document": "shared", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 5, "warmup": null } }, { "adaptive_cell": { "attempts": [], "cell_id": "32x128000-cold", "cell_outcome_sha256": "0f29e25ece1f3cbf766183947da830b2306603371ee572dcee073a211f810b86", "cell_status": "above-confirmed-limit", "limit_reference": { "attempt_sha256": [ "6579ba6d1cc9de0e9fb868caa8e2c9f38d809d86f6c357b89478148b3fb4fc4a", "0ca8f125437dceb4b799ebcfc14a81188debe864b04ced14fceafa36960b3fa3" ], "cell_id": "8x128000-cold", "payload_sha256": "cd9c683c823b57d6165353675fa9eb5f77c0fe6b29b52b61ce6bbed361100a74" }, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "5ef2a2045bd74d3da88c54bd", "licence_class": "commercial", "measured_at": "2026-10-19T01:23:29.636589+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "cold", "concurrency": 32, "configured_context_tokens": 262144, "context_tokens": 128000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 32, "warmup": null } }, { "adaptive_cell": { "attempts": [], "cell_id": "20x250000-warm", "cell_outcome_sha256": "08ab87afe4db6faa08cbbfabc3618c0e105dbf7741aa39dce770de349ec668db", "cell_status": "above-confirmed-limit", "limit_reference": { "attempt_sha256": [ "291e5591b60d97dba4f6913080ce7fab17ad1937b73ac471d6f425205862f2fe", "eb26c7f688d2de53cc76ca52e268e0960055e31e2b8db92a2dabc6da21d75aa1" ], "cell_id": "8x250000-warm", "payload_sha256": "059e2470391458d71c724eaed8cd41b0f975e3544f97606d8f5550381bf372b1" }, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "65a98ee6870f41d89866b7c5", "licence_class": "commercial", "measured_at": "2026-10-19T07:37:30.681102+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "warm", "concurrency": 20, "configured_context_tokens": 262144, "context_tokens": 250000, "scenario": "shared" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": null, "document": "shared", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 20, "warmup": null } }, { "adaptive_cell": { "attempts": [], "cell_id": "64x128000-cold", "cell_outcome_sha256": "c1d63c303bab79c7ea19a55bdc0114a058f5ffe97e73342436722fa844f45640", "cell_status": "above-confirmed-limit", "limit_reference": { "attempt_sha256": [ "6579ba6d1cc9de0e9fb868caa8e2c9f38d809d86f6c357b89478148b3fb4fc4a", "0ca8f125437dceb4b799ebcfc14a81188debe864b04ced14fceafa36960b3fa3" ], "cell_id": "8x128000-cold", "payload_sha256": "cd9c683c823b57d6165353675fa9eb5f77c0fe6b29b52b61ce6bbed361100a74" }, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "6d608ddb95a64b2cb1255da5", "licence_class": "commercial", "measured_at": "2026-10-19T01:23:29.894077+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "cold", "concurrency": 64, "configured_context_tokens": 262144, "context_tokens": 128000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 64, "warmup": null } }, { "adaptive_cell": {}, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "703e57972b734c6ab521195d", "licence_class": "commercial", "measured_at": "2026-10-19T07:37:31.621603+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "configured_context_tokens": 262144 }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "adaptive_cell": { "attempts": [], "cell_id": "16x250000-cold", "cell_outcome_sha256": "6ba30fb5faa535f767e0e7f7f86140f45bc70dabacf525d5f2208b514a16fca0", "cell_status": "above-confirmed-limit", "limit_reference": { "attempt_sha256": [ "e9093e7223960cdffc52b7bb5e2364f5161032d61ed97d7136415d29de1b6bff", "8be552f9423936688b93831ed32f0e11f70f0ba807f2e1d37dc96d6252959f60" ], "cell_id": "8x250000-cold", "payload_sha256": "24c64ed72bb2a1c28206388bf5c8d08a76ff0ee506cfb3191efeaf4053ebe7df" }, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "775f4e21666546a29802c7a6", "licence_class": "commercial", "measured_at": "2026-10-19T04:57:03.743002+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "cold", "concurrency": 16, "configured_context_tokens": 262144, "context_tokens": 250000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 16, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "60d3566d1db44649b1e8cfb2416a200c", "role": "primary", "sha256": "b9165a91a0641300042baee47a526f5f2b41988d683459deb04a06fda7bc26d6", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "2x250000-cold", "cell_outcome_sha256": "7376bf67ed1a6b9f195eac96650cac85608f0f260c2a05c0bfeefd49a99aff9e", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "808dc7e166704db78a291e19", "licence_class": "commercial", "measured_at": "2026-10-19T01:37:28.078968+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 30.95639056145462 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 42.30337731931536 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 78.69805733165283 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 278.7039834150346 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 278.75124783208594 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 41.16911589957272 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 43.43763873905799 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 82.26 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.006321064790227283 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 29583.62336186023 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 34.805880053305025 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 52.76908213927079 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 288.00948575662915 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 288.624413235113 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 26.552216141811947 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 43.0595439647981 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 81.93 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00657210827783392 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 28453.578683526 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 30.300519178353312 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 38.60131131953031 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 277.33390340791084 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 278.5486522628926 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 19.756958276860182 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 40.84408007984644 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 84.93 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.006510279518318825 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 27341.375973049715 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[3].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 30.95639056145462 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[3].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 39.71298217254889 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[3].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 277.3548320890404 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[3].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 278.55785349407233 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[3].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 40.87562595346167 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[3].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 21.03715516944757 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[3].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[3].power_w_median", "statistic": "median", "unit": "W", "value": 84.9 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[3].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.0065157749907681604 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[3].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 28699.579139081678 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[4].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 30.48872815318169 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[4].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 38.737986011886235 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[4].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 277.2833487933967 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[4].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 278.4796915249899 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[4].decode_overlap", "statistic": "single", "unit": "count", "value": 2 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 19.713391764281994 }, { "basis": null, "n": null, "name": "per_person_decode_tps[1]", "scope": "per_person", "source_field": "standard_result.samples[4].per_person_decode_tps[1]", "statistic": "single", "unit": "tok/s", "value": 41.264064542081385 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[1]", "scope": "machine_total", "source_field": "standard_result.samples[4].hidden_facts_per_person[1]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[4].power_w_median", "statistic": "median", "unit": "W", "value": 85.06 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[4].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.006523762893811884 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[4].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 27898.00962457359 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "cold", "concurrency": 2, "configured_context_tokens": 262144, "context_tokens": 250000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 2, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": "observed-latency-boundary", "classification": "capacity", "id": "82a1154119e64aeaaaa853f216aabd49", "role": "primary", "sha256": "291e5591b60d97dba4f6913080ce7fab17ad1937b73ac471d6f425205862f2fe", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, { "capacity_kind": "observed-latency-boundary", "classification": "capacity", "id": "491560e7f2024184a53322bf76bd57e0", "role": "confirmation", "sha256": "eb26c7f688d2de53cc76ca52e268e0960055e31e2b8db92a2dabc6da21d75aa1", "variant_sha256": "ce7a11417103cc42fa9c8145854c46461d126b984c768acf8e243600a467a158" } ], "cell_id": "8x250000-warm", "cell_outcome_sha256": "059e2470391458d71c724eaed8cd41b0f975e3544f97606d8f5550381bf372b1", "cell_status": "capacity-limit-confirmed", "limit_reference": null, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "a461512858134c41a6403f4b", "licence_class": "commercial", "measured_at": "2026-10-19T06:16:24.023031+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "warm", "concurrency": 8, "configured_context_tokens": 262144, "context_tokens": 250000, "scenario": "shared" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": null, "document": "shared", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 8, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": "observed-throughput-boundary", "classification": "capacity", "id": "55d8ec240c464aa9bd3fc6ec08cec9a6", "role": "primary", "sha256": "e9093e7223960cdffc52b7bb5e2364f5161032d61ed97d7136415d29de1b6bff", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, { "capacity_kind": "observed-throughput-boundary", "classification": "capacity", "id": "23433a9ab5084ae1a24a20b9d36896a4", "role": "confirmation", "sha256": "8be552f9423936688b93831ed32f0e11f70f0ba807f2e1d37dc96d6252959f60", "variant_sha256": "ce7a11417103cc42fa9c8145854c46461d126b984c768acf8e243600a467a158" } ], "cell_id": "8x250000-cold", "cell_outcome_sha256": "24c64ed72bb2a1c28206388bf5c8d08a76ff0ee506cfb3191efeaf4053ebe7df", "cell_status": "capacity-limit-confirmed", "limit_reference": null, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "b0872af8db9e49f3bbdbdd7f", "licence_class": "commercial", "measured_at": "2026-10-19T02:47:59.895542+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "cold", "concurrency": 8, "configured_context_tokens": 262144, "context_tokens": 250000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 8, "warmup": null } }, { "adaptive_cell": { "attempts": [], "cell_id": "20x250000-cold", "cell_outcome_sha256": "37e58bffbdc7cb823ea0fd852e87eb5f73082bbfba3e7590572c5807756513d1", "cell_status": "above-confirmed-limit", "limit_reference": { "attempt_sha256": [ "e9093e7223960cdffc52b7bb5e2364f5161032d61ed97d7136415d29de1b6bff", "8be552f9423936688b93831ed32f0e11f70f0ba807f2e1d37dc96d6252959f60" ], "cell_id": "8x250000-cold", "payload_sha256": "24c64ed72bb2a1c28206388bf5c8d08a76ff0ee506cfb3191efeaf4053ebe7df" }, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "b875eee5cafc4c92afd7fa13", "licence_class": "commercial", "measured_at": "2026-10-19T04:57:04.017976+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "cold", "concurrency": 20, "configured_context_tokens": 262144, "context_tokens": 250000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 20, "warmup": null } }, { "adaptive_cell": {}, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "b9a60fe5827e44f69145ff72", "licence_class": "commercial", "measured_at": "2026-10-19T08:54:42.568652+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "configured_context_tokens": 262144 }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "adaptive_cell": {}, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "bc1a11e137ef4f5d8e6dda0a", "licence_class": "commercial", "measured_at": "2026-10-19T07:56:04.833749+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [ "Standard check failed; the retained internal record contains the failure reason." ], "status": "fail" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "configured_context_tokens": 262144 }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "adaptive_cell": {}, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "cdc70e34bcdd4f998702c274", "licence_class": "commercial", "measured_at": "2026-10-19T09:02:33.589780+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "configured_context_tokens": 262144 }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "adaptive_cell": { "attempts": [ { "capacity_kind": null, "classification": "measured", "id": "c877d8687ebb4159a5c333f13af7eb3a", "role": "primary", "sha256": "523adb61fcd5db2d4e1d9f9772168d8ccfd56baea2532d38253003967404a2e2", "variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" } ], "cell_id": "1x250000-warm", "cell_outcome_sha256": "b1c83fdcc31bdca289f0dc4563f2e867ab10e7d720d3c834e35009b4f4c1238e", "cell_status": "measured", "limit_reference": null, "selected_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261" }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "cec07c93bbae4873a4517ec2", "licence_class": "commercial", "measured_at": "2026-10-19T04:57:04.826467+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 61.48775070007293 }, { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[0].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 61.48775070007293 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[0].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 61.48772261599332 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 3.509288491914049 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[0].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 3.509288491914049 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[0].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[0].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 61.48775070007293 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[0].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[0].power_w_median", "statistic": "median", "unit": "W", "value": 14.215 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00016979649201508826 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[0].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 618387.3338836094 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[1].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 61.92497606531819 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[1].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 61.92488424067604 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 3.5269937980920076 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[1].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 3.5269937980920076 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[1].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[1].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 61.92497606531819 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[1].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[1].power_w_median", "statistic": "median", "unit": "W", "value": 14.95 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00017444846366591066 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[1].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 601896.9602454473 }, { "basis": null, "n": null, "name": "median_decode_tps", "scope": "per_person", "source_field": "standard_result.samples[2].median_decode_tps", "statistic": "median", "unit": "tok/s", "value": 61.38217387652993 }, { "basis": null, "n": null, "name": "machine_decode_tps", "scope": "machine_total", "source_field": "standard_result.samples[2].machine_decode_tps", "statistic": "single", "unit": "tok/s", "value": 61.38208034061896 }, { "basis": null, "n": null, "name": "ttft_median_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_median_seconds", "statistic": "median", "unit": "s", "value": 3.492242446169257 }, { "basis": null, "n": null, "name": "ttft_max_seconds", "scope": "per_person", "source_field": "standard_result.samples[2].ttft_max_seconds", "statistic": "max", "unit": "s", "value": 3.492242446169257 }, { "basis": null, "n": null, "name": "decode_overlap", "scope": "machine_total", "source_field": "standard_result.samples[2].decode_overlap", "statistic": "single", "unit": "count", "value": 1 }, { "basis": null, "n": null, "name": "per_person_decode_tps[0]", "scope": "per_person", "source_field": "standard_result.samples[2].per_person_decode_tps[0]", "statistic": "single", "unit": "tok/s", "value": 61.38217387652993 }, { "basis": null, "n": null, "name": "hidden_facts_per_person[0]", "scope": "machine_total", "source_field": "standard_result.samples[2].hidden_facts_per_person[0]", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "power_w_median", "scope": "machine_total", "source_field": "standard_result.samples[2].power_w_median", "statistic": "median", "unit": "W", "value": 15.15 }, { "basis": null, "n": null, "name": "energy_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].energy_kwh", "statistic": "single", "unit": "kWh", "value": 0.00017652942546951493 }, { "basis": null, "n": null, "name": "tokens_per_kwh", "scope": "machine_total", "source_field": "standard_result.samples[2].tokens_per_kwh", "statistic": "single", "unit": "tokens/kWh", "value": 623125.576415565 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "adaptive_variant_sha256": "415588daa437380e860555f3e55a80e6b6c9c4c8dce429d6c940f2fb436ef261", "cache_state": "warm", "concurrency": 1, "configured_context_tokens": 262144, "context_tokens": 250000, "scenario": "shared" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": null, "document": "shared", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 1, "warmup": null } }, { "adaptive_cell": {}, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "d201dcb8efb54cecae49d886", "licence_class": "commercial", "measured_at": "2026-10-19T09:49:10.385938+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [ { "basis": null, "n": null, "name": "n", "scope": "machine_total", "source_field": "standard_result.n", "statistic": "single", "unit": "count", "value": 3 }, { "basis": null, "n": null, "name": "engine_start_to_answer_seconds", "scope": "machine_total", "source_field": "standard_result.samples[0].engine_start_to_answer_seconds", "statistic": "single", "unit": "s", "value": 248.2579635460861 }, { "basis": null, "n": null, "name": "engine_start_to_answer_seconds", "scope": "machine_total", "source_field": "standard_result.samples[1].engine_start_to_answer_seconds", "statistic": "single", "unit": "s", "value": 246.5092204278335 }, { "basis": null, "n": null, "name": "engine_start_to_answer_seconds", "scope": "machine_total", "source_field": "standard_result.samples[2].engine_start_to_answer_seconds", "statistic": "single", "unit": "s", "value": 246.31221690494567 }, { "basis": null, "n": null, "name": "swap_to_answer_seconds", "scope": "machine_total", "source_field": "standard_result.model_swap_samples[0].swap_to_answer_seconds", "statistic": "single", "unit": "s", "value": 422.01062219101004 }, { "basis": null, "n": null, "name": "swap_to_answer_seconds", "scope": "machine_total", "source_field": "standard_result.model_swap_samples[1].swap_to_answer_seconds", "statistic": "single", "unit": "s", "value": 294.26111245388165 }, { "basis": null, "n": null, "name": "swap_to_answer_seconds", "scope": "machine_total", "source_field": "standard_result.model_swap_samples[2].swap_to_answer_seconds", "statistic": "single", "unit": "s", "value": 261.28507626615465 } ], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "pass" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "configured_context_tokens": 262144 }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "adaptive_cell": { "attempts": [], "cell_id": "16x250000-warm", "cell_outcome_sha256": "cd692da2d8d31cc40e76a258fa492ced4a4647b29354a5550a61e14d03b8fb0f", "cell_status": "above-confirmed-limit", "limit_reference": { "attempt_sha256": [ "291e5591b60d97dba4f6913080ce7fab17ad1937b73ac471d6f425205862f2fe", "eb26c7f688d2de53cc76ca52e268e0960055e31e2b8db92a2dabc6da21d75aa1" ], "cell_id": "8x250000-warm", "payload_sha256": "059e2470391458d71c724eaed8cd41b0f975e3544f97606d8f5550381bf372b1" }, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "e263cd1da0b145dbabf8d556", "licence_class": "commercial", "measured_at": "2026-10-19T07:37:30.366316+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "warm", "concurrency": 16, "configured_context_tokens": 262144, "context_tokens": 250000, "scenario": "shared" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": null, "document": "shared", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 16, "warmup": null } }, { "adaptive_cell": { "attempts": [], "cell_id": "32x250000-warm", "cell_outcome_sha256": "8e323b18a4206c0d201b8c170d6cad32067e8148768c22c1ee2bec3be0075e2e", "cell_status": "above-confirmed-limit", "limit_reference": { "attempt_sha256": [ "291e5591b60d97dba4f6913080ce7fab17ad1937b73ac471d6f425205862f2fe", "eb26c7f688d2de53cc76ca52e268e0960055e31e2b8db92a2dabc6da21d75aa1" ], "cell_id": "8x250000-warm", "payload_sha256": "059e2470391458d71c724eaed8cd41b0f975e3544f97606d8f5550381bf372b1" }, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "e48aa24e32834592b8dad808", "licence_class": "commercial", "measured_at": "2026-10-19T07:37:30.991869+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "warm", "concurrency": 32, "configured_context_tokens": 262144, "context_tokens": 250000, "scenario": "shared" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "warm", "content_tokens": null, "document": "shared", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 32, "warmup": null } }, { "adaptive_cell": { "attempts": [], "cell_id": "16x128000-cold", "cell_outcome_sha256": "246d82bda37401c53e794895996029945b944514702cde661edbbb1a6281d51c", "cell_status": "above-confirmed-limit", "limit_reference": { "attempt_sha256": [ "6579ba6d1cc9de0e9fb868caa8e2c9f38d809d86f6c357b89478148b3fb4fc4a", "0ca8f125437dceb4b799ebcfc14a81188debe864b04ced14fceafa36960b3fa3" ], "cell_id": "8x128000-cold", "payload_sha256": "cd9c683c823b57d6165353675fa9eb5f77c0fe6b29b52b61ce6bbed361100a74" }, "selected_variant_sha256": null }, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "e8547ed4f1614104bc31aae7", "licence_class": "commercial", "measured_at": "2026-10-19T01:23:29.138868+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [], "status": "not_graded" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "cache_state": "cold", "concurrency": 16, "configured_context_tokens": 262144, "context_tokens": 128000, "scenario": "distinct" }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": "cold", "content_tokens": null, "document": "distinct", "input_tokens": null, "output_tokens": null, "repetitions": null, "users": 16, "warmup": null } }, { "adaptive_cell": {}, "caveats": [ "Labelled lab evidence for the recorded setup and standard. Completion is coverage, not customer or general answer-quality qualification.", "Publisher revision was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "plain", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.2", "history": { "corrections": [], "supersedes": [] }, "id": "fac9a72d5bce40bcbcacb31e", "licence_class": "commercial", "measured_at": "2026-10-19T08:01:15.573692+00:00", "methodology": { "implementation_sha256": "2de3c503cd76f5fc9ce6d5148760c219ddb52ee5cfbf5d728c3697037536d047", "protocol_sha256": "25fe3b0656961091ac1dfcc41df0b8cce189498814bf6c134c310f95ac15d1b0", "revision": "KTS-1/adaptive-capacity-v2" }, "metrics": [], "model": { "artifact_sha256": [ "07141c2db92e47bc08777132cd1a0323faf300eab3a7d7c111bc2bf075fda050", "6dea9c759a0f941cf9e1cc1501216b0e107966a66c90425050e212efbd053f02", "9758875fc55e49561165f4a44342b654c123c3e25c6811b34abb83553fb1a164" ], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": null, "variant": "NVFP4 variant", "weights_dtype": "nvfp4", "weights_format": "safetensors" }, "outcome": { "gates": [], "reasons": [ "Standard check failed; the retained internal record contains the failure reason." ], "status": "fail" }, "publication_group": "results", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [], "record_date": "2026-10-19", "reproduce": { "driver": null, "fixture_sha256": null, "harness": { "name": null, "version": null }, "harness_sha256": null, "os": null, "workload_name": null, "workload_version": null }, "requests": [], "scenario": { "id": "test-standard", "version": null }, "settings": { "configured_context_tokens": 262144 }, "setup_complete": true, "setup_id": "4de08bcd661f3059c9f0cd0f7904c9ed", "standard": "KTS-1", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": null, "repetitions": null, "users": null, "warmup": null } }, { "caveats": [ "Lab workload, not a KTS-1 block and not customer acceptance. One repository and one fictional shop; Keep's own minimal agent loop, not a product coding harness. The model may have seen this open-source repository in training; the planted bugs and shop data are new. Engines do not guarantee identical output under concurrent batching. Original publisher name with variant; exact variant repository and revision retained privately.", "Weights file SHA-256 was not recorded for public reproduction." ], "engine": { "name": "NVIDIA NGC vLLM", "recipe": "with multi-token prediction", "version": "0.29.0+5013de39.dev" }, "format": "olir/0.1", "history": { "corrections": [], "supersedes": [] }, "id": "1bdb1c2c257c141430090354", "measured_at": "2026-10-21T08:07:59.563191+00:00", "metrics": [ { "basis": null, "n": null, "name": "tasks_passed", "scope": "machine_total", "source_field": "agent_team.tasks_passed", "statistic": "single", "unit": "count", "value": 38 }, { "basis": null, "n": null, "name": "tasks_total", "scope": "machine_total", "source_field": "agent_team.tasks_total", "statistic": "single", "unit": "count", "value": 40 }, { "basis": null, "n": null, "name": "tool_calls", "scope": "machine_total", "source_field": "agent_team.tool_calls", "statistic": "single", "unit": "count", "value": 702 }, { "basis": null, "n": null, "name": "protocol_errors", "scope": "machine_total", "source_field": "agent_team.protocol_errors", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "engine_errors", "scope": "machine_total", "source_field": "agent_team.engine_errors", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "loops", "scope": "machine_total", "source_field": "agent_team.loops", "statistic": "single", "unit": "count", "value": 0 }, { "basis": null, "n": null, "name": "model_requests", "scope": "machine_total", "source_field": "agent_team.model_requests", "statistic": "single", "unit": "count", "value": 599 }, { "basis": null, "n": null, "name": "peak_concurrent_requests", "scope": "machine_total", "source_field": "agent_team.peak_concurrent_requests", "statistic": "max", "unit": "count", "value": 5 }, { "basis": null, "n": null, "name": "mean_concurrent_requests", "scope": "machine_total", "source_field": "agent_team.mean_concurrent_requests", "statistic": "mean", "unit": "count", "value": 2.9560184109560557 }, { "basis": null, "n": null, "name": "ttft_p50_s", "scope": "per_person", "source_field": "agent_team.ttft_p50_s", "statistic": "median", "unit": "s", "value": 1.019965103827417 }, { "basis": null, "n": null, "name": "ttft_p95_s", "scope": "per_person", "source_field": "agent_team.ttft_p95_s", "statistic": "p95", "unit": "s", "value": 2.2109051721636206 }, { "basis": null, "n": null, "name": "per_request_decode_p50_tps", "scope": "per_person", "source_field": "agent_team.per_request_decode_p50_tps", "statistic": "median", "unit": "tok/s", "value": 42.89717956971035 }, { "basis": null, "n": null, "name": "task_p50_seconds", "scope": "per_person", "source_field": "agent_team.task_p50_seconds", "statistic": "median", "unit": "s", "value": 66.92939303885214 }, { "basis": null, "n": null, "name": "task_p95_seconds", "scope": "per_person", "source_field": "agent_team.task_p95_seconds", "statistic": "p95", "unit": "s", "value": 656.0206727720797 }, { "basis": null, "n": null, "name": "max_prompt_tokens", "scope": "machine_total", "source_field": "agent_team.max_prompt_tokens", "statistic": "max", "unit": "tokens", "value": 51639 }, { "basis": null, "n": null, "name": "wall_seconds", "scope": "machine_total", "source_field": "agent_team.wall_seconds", "statistic": "single", "unit": "s", "value": 2083.3306227349676 } ], "model": { "artifact_sha256": [], "publisher_repo": "Qwen/Qwen3.6-35B-A3B", "publisher_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "variant": "NVFP4 variant", "weights_dtype": "NVFP4", "weights_format": "safetensors" }, "outcome": { "gates": [ { "met": true, "name": "agent-pass-rate", "observed": null }, { "met": true, "name": "agent-category-pass-rate", "observed": null }, { "met": true, "name": "agent-protocol-error-rate", "observed": null }, { "met": true, "name": "agent-engine-errors", "observed": null }, { "met": true, "name": "agent-loops", "observed": null }, { "met": true, "name": "agent-concurrency", "observed": null }, { "met": true, "name": "agent-complete", "observed": null }, { "met": true, "name": "agent-ttft-p95", "observed": null }, { "met": true, "name": "agent-decode-p50", "observed": null }, { "met": true, "name": "agent-task-seconds-p95", "observed": null } ], "reasons": [], "status": "pass" }, "publication_group": "lab workload", "published": null, "publisher": { "name": "Skilak Keep" }, "quality": [ { "passed": 38, "skipped": null, "suite": "agent-tasks", "total": 40, "version": "agent-team-v1" }, { "passed": 10, "skipped": null, "suite": "agent-tasks-bugfix", "total": 12, "version": "agent-team-v1" }, { "passed": 12, "skipped": null, "suite": "agent-tasks-data", "total": 12, "version": "agent-team-v1" }, { "passed": 6, "skipped": null, "suite": "agent-tasks-feature", "total": 6, "version": "agent-team-v1" }, { "passed": 8, "skipped": null, "suite": "agent-tasks-investigate", "total": 8, "version": "agent-team-v1" }, { "passed": 2, "skipped": null, "suite": "agent-tasks-multistep", "total": 2, "version": "agent-team-v1" } ], "record_date": "2026-10-21", "reproduce": { "driver": null, "fixture_sha256": "6451aabd1c9994d0f73355c962a5eaf9258381b3f2554e53db916cd9a605552a", "harness": { "name": "keep-agent-team-result-v1", "version": "1" }, "harness_sha256": "484ed3e2b82f09ee19c87be819253811cb273628abfb3c9471a59746605e537d", "os": null, "workload_name": "agent-team-v1", "workload_version": "1" }, "requests": [], "scenario": { "id": "agent-team", "version": "1" }, "settings": { "concurrency": 5, "configured_context_tokens": 262144, "kv_dtype": "fp8", "max_output_tokens": 8192, "speculative_decoding": "with multi-token prediction" }, "setup_complete": false, "setup_id": "1613bbe6cf3a8a03e019ade665b34366", "standard": "lab workload", "system": { "driver": null, "hardware": "NVIDIA GB10, 128 GB unified memory", "os": null }, "workload": { "cache": null, "content_tokens": null, "document": null, "input_tokens": null, "output_tokens": 8192, "repetitions": null, "users": 5, "warmup": null } } ]