{ "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 }, { "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 }, { "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 }, { "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 }, { "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 }, { "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 }, { "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 }, { "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 }, { "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 }, { "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 }, { "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 }, { "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 }, { "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 }, { "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 }, { "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 }, { "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 }, { "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 }, { "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 }, { "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 }, { "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 }, { "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 } }