{ "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 }, { "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 } }