{
  "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
  }
}