{ "base_url": "http://127.0.0.1:8001/v1", "clock_domain_id": null, "completed": 60, "concurrency": 1, "configuration_fingerprint": null, "context_distribution": { "4096": 60 }, "context_seed": 0, "context_tokens": 4096, "controlled_output_policy": "observe", "engine": "sglang-glm53-sm120-v0.1.1-rc14", "failed": 0, "failures": [], "finished_at": "2026-09-08T05:30:33Z", "finished_at_unix_ns": 1788845433035761941, "gpu": "2x RTX PRO 6000 Blackwell Max-Q TP2 Linux", "manifest": null, "max_context_tokens": 393216, "max_tokens": 256, "measurement_protocol": "capacity-v3", "metric_population": { "aggregate_throughput_eligible": true, "attempted_requests": 60, "excluded_requests": 0, "failure_policy": "failed or contract-invalid responses retain sanitized timing and metadata but are excluded from performance distributions", "p99_interpretation": "descriptive-nearest-rank-only", "p99_interpretation_by_metric": { "decode_tok_s": "descriptive-nearest-rank-only", "e2e": "descriptive-nearest-rank-only", "effective_prefill_tok_s": "descriptive-nearest-rank-only", "tpot": "descriptive-nearest-rank-only", "ttft": "descriptive-nearest-rank-only" }, "p99_minimum_recommended_samples": 100, "p99_samples_by_metric": { "decode_tok_s": 60, "e2e": 60, "effective_prefill_tok_s": 60, "tpot": 60, "ttft": 60 }, "performance_eligible_requests": 60, "successful_requests": 60 }, "metrics": { "completion_tokens_max": 256.0, "completion_tokens_mean": 74.4, "completion_tokens_mean_ci95_high": 91.06414876890177, "completion_tokens_mean_ci95_low": 57.73585123109824, "completion_tokens_min": 5.0, "completion_tokens_p25": 36.0, "completion_tokens_p50": 52.0, "completion_tokens_p75": 84.0, "completion_tokens_p90": 183.0, "completion_tokens_p95": 234.0, "completion_tokens_p99": 256.0, "completion_tokens_samples": 60, "completion_tokens_stddev": 65.8571129190606, "content_chunks": 1812, "content_chunks_s": null, "controlled_output_exact_adherence": 0, "controlled_output_exact_adherence_rate": null, "controlled_output_samples": 0, "controlled_output_unobservable": 0, "decode_tok_s_max": 172.00882231721636, "decode_tok_s_mean": 136.37353030590978, "decode_tok_s_mean_ci95_high": 143.15220324978904, "decode_tok_s_mean_ci95_low": 129.59485736203052, "decode_tok_s_min": 60.79870960692588, "decode_tok_s_p25": 127.87753900005823, "decode_tok_s_p50": 142.6429205479183, "decode_tok_s_p75": 152.85283942663446, "decode_tok_s_p90": 162.1677214039653, "decode_tok_s_p95": 164.13184246288014, "decode_tok_s_p99": 172.00882231721636, "decode_tok_s_samples": 60, "decode_tok_s_stddev": 26.789476960235852, "e2e_max_ms": 2047.0468740022625, "e2e_mean_ci95_high_ms": 743.2945546814049, "e2e_mean_ci95_low_ms": 523.7451468178535, "e2e_mean_ms": 633.5198507496292, "e2e_min_ms": 169.48487799527356, "e2e_p25_ms": 377.29013299394865, "e2e_p50_ms": 477.16181799478363, "e2e_p75_ms": 666.8558429955738, "e2e_p90_ms": 1179.5730000012554, "e2e_p95_ms": 1649.7575439934735, "e2e_p99_ms": 2047.0468740022625, "e2e_samples": 60, "e2e_stddev_ms": 433.83224506389513, "effective_prefill_tok_s_max": 25155.33248260817, "effective_prefill_tok_s_mean": 23773.130508048816, "effective_prefill_tok_s_mean_ci95_high": 24398.87367255708, "effective_prefill_tok_s_mean_ci95_low": 23147.38734354055, "effective_prefill_tok_s_min": 5508.1202428848055, "effective_prefill_tok_s_p25": 23672.645503661624, "effective_prefill_tok_s_p50": 24051.258651905544, "effective_prefill_tok_s_p75": 24447.23698025414, "effective_prefill_tok_s_p90": 24876.10751695148, "effective_prefill_tok_s_p95": 24994.68831249027, "effective_prefill_tok_s_p99": 25155.33248260817, "effective_prefill_tok_s_samples": 60, "effective_prefill_tok_s_stddev": 2472.951893003716, "generation_max_ms": 1924.3150530019193, "generation_mean_ci95_high_ms": 612.418298441515, "generation_mean_ci95_low_ms": 394.0422941920542, "generation_mean_ms": 503.23029631678463, "generation_min_ms": 49.203703994862735, "generation_p25_ms": 252.82466600037878, "generation_p50_ms": 355.85797100065975, "generation_p75_ms": 528.0302180035505, "generation_p90_ms": 1058.0852630009758, "generation_p95_ms": 1531.357237996417, "generation_p99_ms": 1924.3150530019193, "generation_samples": 60, "generation_stddev_ms": 431.51358554565417, "mean_inter_token_latency_ms_max": 16.44771750034124, "mean_inter_token_latency_ms_mean": 7.789553993125394, "mean_inter_token_latency_ms_mean_ci95_high": 8.412366297042993, "mean_inter_token_latency_ms_mean_ci95_low": 7.166741689207795, "mean_inter_token_latency_ms_min": 5.813655291214153, "mean_inter_token_latency_ms_p25": 6.490535690913517, "mean_inter_token_latency_ms_p50": 6.97760727452274, "mean_inter_token_latency_ms_p75": 7.716187896409698, "mean_inter_token_latency_ms_p90": 11.027758665780615, "mean_inter_token_latency_ms_p95": 12.339291499301908, "mean_inter_token_latency_ms_p99": 16.44771750034124, "mean_inter_token_latency_ms_samples": 60, "mean_inter_token_latency_ms_stddev": 2.4613690621285036, "mean_time_per_output_token_max_ms": 16.44771750034124, "mean_time_per_output_token_mean_ci95_high_ms": 8.412366297042993, "mean_time_per_output_token_mean_ci95_low_ms": 7.166741689207795, "mean_time_per_output_token_mean_ms": 7.789553993125394, "mean_time_per_output_token_min_ms": 5.813655291214153, "mean_time_per_output_token_p25_ms": 6.490535690913517, "mean_time_per_output_token_p50_ms": 6.97760727452274, "mean_time_per_output_token_p75_ms": 7.716187896409698, "mean_time_per_output_token_p90_ms": 11.027758665780615, "mean_time_per_output_token_p95_ms": 12.339291499301908, "mean_time_per_output_token_p99_ms": 16.44771750034124, "mean_time_per_output_token_samples": 60, "mean_time_per_output_token_stddev_ms": 2.4613690621285036, "output_token_sources": { "usage": 60 }, "output_tokens": 4464, "prefix_cache_hit_avg": 0.9915796564499831, "prompt_token_samples": 60, "prompt_tokens": 178140, "prompt_tokens_max": 2969.0, "prompt_tokens_mean": 2969.0, "prompt_tokens_mean_ci95_high": 2969.0, "prompt_tokens_mean_ci95_low": 2969.0, "prompt_tokens_min": 2969.0, "prompt_tokens_p25": 2969.0, "prompt_tokens_p50": 2969.0, "prompt_tokens_p75": 2969.0, "prompt_tokens_p90": 2969.0, "prompt_tokens_p95": 2969.0, "prompt_tokens_p99": 2969.0, "prompt_tokens_samples": 60, "prompt_tokens_stddev": 0.0, "reasoning_chunks": 0, "request_canary_passed": 0, "request_canary_samples": 0, "throughput_performance_eligible": true, "throughput_tok_s": 117.41657365735207, "time_to_first_output_max_ms": 539.022364995617, "time_to_first_output_mean_ci95_high_ms": 143.89081019066694, "time_to_first_output_mean_ci95_low_ms": 116.68829867502214, "time_to_first_output_mean_ms": 130.28955443284454, "time_to_first_output_min_ms": 118.02666500443593, "time_to_first_output_p25_ms": 121.38976699498016, "time_to_first_output_p50_ms": 123.24580299900845, "time_to_first_output_p75_ms": 125.15283899847418, "time_to_first_output_p90_ms": 127.85984799847938, "time_to_first_output_p95_ms": 129.3051429965999, "time_to_first_output_p99_ms": 539.022364995617, "time_to_first_output_samples": 60, "time_to_first_output_stddev_ms": 53.75248677301417, "tpot_max_ms": 16.44771750034124, "tpot_mean_ci95_high_ms": 8.412366297042993, "tpot_mean_ci95_low_ms": 7.166741689207795, "tpot_mean_ms": 7.789553993125394, "tpot_min_ms": 5.813655291214153, "tpot_p25_ms": 6.490535690913517, "tpot_p50_ms": 6.97760727452274, "tpot_p75_ms": 7.716187896409698, "tpot_p90_ms": 11.027758665780615, "tpot_p95_ms": 12.339291499301908, "tpot_p99_ms": 16.44771750034124, "tpot_samples": 60, "tpot_stddev_ms": 2.4613690621285036, "ttft_max_ms": 539.0233349971822, "ttft_mean_ci95_high_ms": 143.8912406219385, "ttft_mean_ci95_low_ms": 116.6886922452228, "ttft_mean_ms": 130.28996643358064, "ttft_min_ms": 118.02710500342073, "ttft_p25_ms": 121.39022699557245, "ttft_p50_ms": 123.24624299799325, "ttft_p75_ms": 125.1532189999125, "ttft_p90_ms": 127.86016799509525, "ttft_p95_ms": 129.3055229980382, "ttft_p99_ms": 539.0233349971822, "ttft_samples": 60, "ttft_stddev_ms": 53.75255961092933 }, "model": "glm53-flash-ormandj-sglang-sm120-tp2-393k-c1-adaptive-mtp", "performance_eligible": true, "prompt_cache_mode": "shared", "request_canaries": false, "request_timings": [ { "content_chunks": 29, "controlled_output": null, "decode_tok_s": 138.66299426569657, "e2e_ms": 1014.9965089949546, "effective_prefill_tok_s": 5508.1202428848055, "generation_ms": 475.9741439993377, "mean_inter_token_latency_ms": 7.211729454535419, "mean_time_per_output_token_ms": 7.211729454535419, "output_token_source": "usage", "output_tokens": 67, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 0, "time_to_first_output_ms": 539.022364995617, "tpot_ms": 7.211729454535419, "ttft_ms": 539.0233349971822, "visible_content_truncated": false, "visible_generation_ms": 475.9731739977724 }, { "content_chunks": 21, "controlled_output": null, "decode_tok_s": 154.6174675161256, "e2e_ms": 477.16181799478363, "effective_prefill_tok_s": 24447.23698025414, "generation_ms": 355.71660099958535, "mean_inter_token_latency_ms": 6.467574563628824, "mean_time_per_output_token_ms": 6.467574563628824, "output_token_source": "usage", "output_tokens": 56, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 1, "time_to_first_output_ms": 121.44521699519828, "tpot_ms": 6.467574563628824, "ttft_ms": 121.44571699900553, "visible_content_truncated": false, "visible_generation_ms": 355.7161009957781 }, { "content_chunks": 98, "controlled_output": null, "decode_tok_s": 154.31726952652863, "e2e_ms": 1778.4352240050794, "effective_prefill_tok_s": 23564.35029224703, "generation_ms": 1652.4398130059126, "mean_inter_token_latency_ms": 6.480156129434952, "mean_time_per_output_token_ms": 6.480156129434952, "output_token_source": "usage", "output_tokens": 256, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 2, "time_to_first_output_ms": 125.99541099916678, "tpot_ms": 6.480156129434952, "ttft_ms": 125.99567100551212, "visible_content_truncated": false, "visible_generation_ms": 1652.4395529995672 }, { "content_chunks": 20, "controlled_output": null, "decode_tok_s": 138.8523855637143, "e2e_ms": 462.0927370051504, "effective_prefill_tok_s": 24020.302162911612, "generation_ms": 338.48896300332854, "mean_inter_token_latency_ms": 7.201892829858054, "mean_time_per_output_token_ms": 7.201892829858054, "output_token_source": "usage", "output_tokens": 48, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 3, "time_to_first_output_ms": 123.60377400182188, "tpot_ms": 7.201892829858054, "ttft_ms": 123.60390400135657, "visible_content_truncated": false, "visible_generation_ms": 338.48883300379384 }, { "content_chunks": 20, "controlled_output": null, "decode_tok_s": 138.98920755433986, "e2e_ms": 464.62450599938165, "effective_prefill_tok_s": 23476.154433713, "generation_ms": 338.1557519969647, "mean_inter_token_latency_ms": 7.194803233977972, "mean_time_per_output_token_ms": 7.194803233977972, "output_token_source": "usage", "output_tokens": 48, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 4, "time_to_first_output_ms": 126.46875400241697, "tpot_ms": 7.194803233977972, "ttft_ms": 126.46902399865212, "visible_content_truncated": false, "visible_generation_ms": 338.15548200072953 }, { "content_chunks": 16, "controlled_output": null, "decode_tok_s": 150.6127930595945, "e2e_ms": 386.39774399780435, "effective_prefill_tok_s": 23294.386985382953, "generation_ms": 258.94214699656004, "mean_inter_token_latency_ms": 6.639542230681027, "mean_time_per_output_token_ms": 6.639542230681027, "output_token_source": "usage", "output_tokens": 40, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 5, "time_to_first_output_ms": 127.45559700124431, "tpot_ms": 6.639542230681027, "ttft_ms": 127.45599700429011, "visible_content_truncated": false, "visible_generation_ms": 258.94174699351424 }, { "content_chunks": 21, "controlled_output": null, "decode_tok_s": 140.30860029165214, "e2e_ms": 484.35353099921485, "effective_prefill_tok_s": 23196.00330613876, "generation_ms": 356.357343000127, "mean_inter_token_latency_ms": 7.12714686000254, "mean_time_per_output_token_ms": 7.12714686000254, "output_token_source": "usage", "output_tokens": 51, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 6, "time_to_first_output_ms": 127.99618799908785, "tpot_ms": 7.12714686000254, "ttft_ms": 127.99667799845338, "visible_content_truncated": false, "visible_generation_ms": 356.3568530007615 }, { "content_chunks": 16, "controlled_output": null, "decode_tok_s": 157.14689626789766, "e2e_ms": 382.3987599971588, "effective_prefill_tok_s": 23220.73775682347, "generation_ms": 254.5389119986794, "mean_inter_token_latency_ms": 6.363472799966985, "mean_time_per_output_token_ms": 6.363472799966985, "output_token_source": "usage", "output_tokens": 41, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 7, "time_to_first_output_ms": 127.85984799847938, "tpot_ms": 6.363472799966985, "ttft_ms": 127.86016799509525, "visible_content_truncated": false, "visible_generation_ms": 254.53859200206352 }, { "content_chunks": 16, "controlled_output": null, "decode_tok_s": 150.9472257728176, "e2e_ms": 389.16374299878953, "effective_prefill_tok_s": 22699.592763873366, "generation_ms": 258.3684450000874, "mean_inter_token_latency_ms": 6.624831923079164, "mean_time_per_output_token_ms": 6.624831923079164, "output_token_source": "usage", "output_tokens": 40, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 8, "time_to_first_output_ms": 130.79529799870215, "tpot_ms": 6.624831923079164, "ttft_ms": 130.795657998533, "visible_content_truncated": false, "visible_generation_ms": 258.36808500025654 }, { "content_chunks": 29, "controlled_output": null, "decode_tok_s": 152.00765503751668, "e2e_ms": 596.9061590003548, "effective_prefill_tok_s": 24090.06982593871, "generation_ms": 473.6603560013464, "mean_inter_token_latency_ms": 6.578616055574255, "mean_time_per_output_token_ms": 6.578616055574255, "output_token_source": "usage", "output_tokens": 73, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 9, "time_to_first_output_ms": 123.24580299900845, "tpot_ms": 6.578616055574255, "ttft_ms": 123.24624299799325, "visible_content_truncated": false, "visible_generation_ms": 473.65991600236157 }, { "content_chunks": 26, "controlled_output": null, "decode_tok_s": 150.12622817816825, "e2e_ms": 566.7661680054152, "effective_prefill_tok_s": 23352.921733568808, "generation_ms": 439.63004200486466, "mean_inter_token_latency_ms": 6.66106124249795, "mean_time_per_output_token_ms": 6.66106124249795, "output_token_source": "usage", "output_tokens": 67, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 10, "time_to_first_output_ms": 127.13612600055058, "tpot_ms": 6.66106124249795, "ttft_ms": 127.13647600321565, "visible_content_truncated": false, "visible_generation_ms": 439.6296920021996 }, { "content_chunks": 29, "controlled_output": null, "decode_tok_s": 152.85283942663446, "e2e_ms": 606.1221700001624, "effective_prefill_tok_s": 23098.11437202177, "generation_ms": 477.5835390028078, "mean_inter_token_latency_ms": 6.542240260312436, "mean_time_per_output_token_ms": 6.542240260312436, "output_token_source": "usage", "output_tokens": 74, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 11, "time_to_first_output_ms": 128.53863099735463, "tpot_ms": 6.542240260312436, "ttft_ms": 128.5391810015426, "visible_content_truncated": false, "visible_generation_ms": 477.58298899861984 }, { "content_chunks": 60, "controlled_output": null, "decode_tok_s": 172.00882231721636, "e2e_ms": 1179.5730000012554, "effective_prefill_tok_s": 24438.680588750838, "generation_ms": 1058.0852630009758, "mean_inter_token_latency_ms": 5.813655291214153, "mean_time_per_output_token_ms": 5.813655291214153, "output_token_source": "usage", "output_tokens": 183, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 12, "time_to_first_output_ms": 121.4877370002796, "tpot_ms": 5.813655291214153, "ttft_ms": 121.48813700332539, "visible_content_truncated": false, "visible_generation_ms": 1058.08486299793 }, { "content_chunks": 14, "controlled_output": null, "decode_tok_s": 157.4408092555487, "e2e_ms": 372.7656780029065, "effective_prefill_tok_s": 22594.247271842178, "generation_ms": 241.36054800328566, "mean_inter_token_latency_ms": 6.351593368507517, "mean_time_per_output_token_ms": 6.351593368507517, "output_token_source": "usage", "output_tokens": 39, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 13, "time_to_first_output_ms": 131.40512999962084, "tpot_ms": 6.351593368507517, "ttft_ms": 131.4055000038934, "visible_content_truncated": false, "visible_generation_ms": 241.3601779990131 }, { "content_chunks": 33, "controlled_output": null, "decode_tok_s": 164.13184246288014, "e2e_ms": 667.6660250013811, "effective_prefill_tok_s": 23672.645503661624, "generation_ms": 542.2470049961703, "mean_inter_token_latency_ms": 6.09266297748506, "mean_time_per_output_token_ms": 6.09266297748506, "output_token_source": "usage", "output_tokens": 90, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 14, "time_to_first_output_ms": 125.41902000521077, "tpot_ms": 6.09266297748506, "ttft_ms": 125.41929000144592, "visible_content_truncated": false, "visible_generation_ms": 542.2467349999351 }, { "content_chunks": 20, "controlled_output": null, "decode_tok_s": 138.79072480072688, "e2e_ms": 462.79363900248427, "effective_prefill_tok_s": 23913.792108235884, "generation_ms": 338.6393440014217, "mean_inter_token_latency_ms": 7.205092425562164, "mean_time_per_output_token_ms": 7.205092425562164, "output_token_source": "usage", "output_tokens": 48, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 15, "time_to_first_output_ms": 124.15429500106256, "tpot_ms": 7.205092425562164, "ttft_ms": 124.15482500364305, "visible_content_truncated": false, "visible_generation_ms": 338.6388139988412 }, { "content_chunks": 15, "controlled_output": null, "decode_tok_s": 166.52954069198196, "e2e_ms": 365.05586199928075, "effective_prefill_tok_s": 23778.967632183616, "generation_ms": 240.19762399984756, "mean_inter_token_latency_ms": 6.004940599996189, "mean_time_per_output_token_ms": 6.004940599996189, "output_token_source": "usage", "output_tokens": 41, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 16, "time_to_first_output_ms": 124.85823799943319, "tpot_ms": 6.004940599996189, "ttft_ms": 124.85870799719123, "visible_content_truncated": false, "visible_generation_ms": 240.19715400208952 }, { "content_chunks": 91, "controlled_output": null, "decode_tok_s": 152.15260960587509, "e2e_ms": 1649.7575439934735, "effective_prefill_tok_s": 25075.948706364095, "generation_ms": 1531.357237996417, "mean_inter_token_latency_ms": 6.572348660928829, "mean_time_per_output_token_ms": 6.572348660928829, "output_token_source": "usage", "output_tokens": 234, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 17, "time_to_first_output_ms": 118.40030599705642, "tpot_ms": 6.572348660928829, "ttft_ms": 118.40061599650653, "visible_content_truncated": false, "visible_generation_ms": 1531.356927996967 }, { "content_chunks": 31, "controlled_output": null, "decode_tok_s": 142.6429205479183, "e2e_ms": 634.8927159997402, "effective_prefill_tok_s": 24113.64839562987, "generation_ms": 511.7674239954795, "mean_inter_token_latency_ms": 7.010512657472321, "mean_time_per_output_token_ms": 7.010512657472321, "output_token_source": "usage", "output_tokens": 74, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 18, "time_to_first_output_ms": 123.1252920042607, "tpot_ms": 7.010512657472321, "ttft_ms": 123.12581200239947, "visible_content_truncated": false, "visible_generation_ms": 511.7669039973407 }, { "content_chunks": 20, "controlled_output": null, "decode_tok_s": 150.24370987968973, "e2e_ms": 461.8803669945919, "effective_prefill_tok_s": 24250.219796860336, "generation_ms": 339.4484870004817, "mean_inter_token_latency_ms": 6.655852686283954, "mean_time_per_output_token_ms": 6.655852686283954, "output_token_source": "usage", "output_tokens": 52, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 19, "time_to_first_output_ms": 122.4318799941102, "tpot_ms": 6.655852686283954, "ttft_ms": 122.43218999356031, "visible_content_truncated": false, "visible_generation_ms": 339.4481770010316 }, { "content_chunks": 16, "controlled_output": null, "decode_tok_s": 162.1677214039653, "e2e_ms": 377.29013299394865, "effective_prefill_tok_s": 23854.006028462372, "generation_ms": 252.82466600037878, "mean_inter_token_latency_ms": 6.166455268301921, "mean_time_per_output_token_ms": 6.166455268301921, "output_token_source": "usage", "output_tokens": 42, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 20, "time_to_first_output_ms": 124.46546699356986, "tpot_ms": 6.166455268301921, "ttft_ms": 124.46575699868845, "visible_content_truncated": false, "visible_generation_ms": 252.8243759952602 }, { "content_chunks": 15, "controlled_output": null, "decode_tok_s": 143.70683548185508, "e2e_ms": 361.7456210049568, "effective_prefill_tok_s": 23722.993611325084, "generation_ms": 236.5927820064826, "mean_inter_token_latency_ms": 6.958611235484782, "mean_time_per_output_token_ms": 6.958611235484782, "output_token_source": "usage", "output_tokens": 35, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 21, "time_to_first_output_ms": 125.15283899847418, "tpot_ms": 6.958611235484782, "ttft_ms": 125.1532189999125, "visible_content_truncated": false, "visible_generation_ms": 236.59240200504428 }, { "content_chunks": 53, "controlled_output": null, "decode_tok_s": 127.87753900005823, "e2e_ms": 1007.847494998714, "effective_prefill_tok_s": 23906.99844324611, "generation_ms": 883.6579190028715, "mean_inter_token_latency_ms": 7.819981584096208, "mean_time_per_output_token_ms": 7.819981584096208, "output_token_source": "usage", "output_tokens": 114, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 22, "time_to_first_output_ms": 124.18957599584246, "tpot_ms": 7.819981584096208, "ttft_ms": 124.18977599736536, "visible_content_truncated": false, "visible_generation_ms": 883.6577190013486 }, { "content_chunks": 84, "controlled_output": null, "decode_tok_s": 154.0704878027185, "e2e_ms": 1551.856817000953, "effective_prefill_tok_s": 23955.339630280876, "generation_ms": 1427.9178520009737, "mean_inter_token_latency_ms": 6.490535690913517, "mean_time_per_output_token_ms": 6.490535690913517, "output_token_source": "usage", "output_tokens": 221, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 23, "time_to_first_output_ms": 123.93896499997936, "tpot_ms": 6.490535690913517, "ttft_ms": 123.93947500095237, "visible_content_truncated": false, "visible_generation_ms": 1427.9173420000006 }, { "content_chunks": 4, "controlled_output": null, "decode_tok_s": 60.79870960692588, "e2e_ms": 188.01144899771316, "effective_prefill_tok_s": 24292.144779388665, "generation_ms": 65.79087000136496, "mean_inter_token_latency_ms": 16.44771750034124, "mean_time_per_output_token_ms": 16.44771750034124, "output_token_source": "usage", "output_tokens": 5, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 24, "time_to_first_output_ms": 122.2205789963482, "tpot_ms": 16.44771750034124, "ttft_ms": 122.22109900176292, "visible_content_truncated": false, "visible_generation_ms": 65.79034999595024 }, { "content_chunks": 96, "controlled_output": null, "decode_tok_s": 157.91323107282568, "e2e_ms": 1736.36959400028, "effective_prefill_tok_s": 24424.406449198614, "generation_ms": 1614.8108569977921, "mean_inter_token_latency_ms": 6.332591596069773, "mean_time_per_output_token_ms": 6.332591596069773, "output_token_source": "usage", "output_tokens": 256, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 25, "time_to_first_output_ms": 121.55873700248776, "tpot_ms": 6.332591596069773, "ttft_ms": 121.55905699910363, "visible_content_truncated": false, "visible_generation_ms": 1614.8105370011763 }, { "content_chunks": 17, "controlled_output": null, "decode_tok_s": 121.37799606874388, "e2e_ms": 409.85640199505724, "effective_prefill_tok_s": 24436.011464313407, "generation_ms": 288.35539499414153, "mean_inter_token_latency_ms": 8.238725571261186, "mean_time_per_output_token_ms": 8.238725571261186, "output_token_source": "usage", "output_tokens": 36, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 26, "time_to_first_output_ms": 121.5010070009157, "tpot_ms": 8.238725571261186, "ttft_ms": 121.50132699753158, "visible_content_truncated": false, "visible_generation_ms": 288.35507499752566 }, { "content_chunks": 22, "controlled_output": null, "decode_tok_s": 112.84630940343922, "e2e_ms": 495.63230900093913, "effective_prefill_tok_s": 24051.258651905544, "generation_ms": 372.1876260024146, "mean_inter_token_latency_ms": 8.861610142914634, "mean_time_per_output_token_ms": 8.861610142914634, "output_token_source": "usage", "output_tokens": 43, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 27, "time_to_first_output_ms": 123.44468299852451, "tpot_ms": 8.861610142914634, "ttft_ms": 123.44503300118959, "visible_content_truncated": false, "visible_generation_ms": 372.18727599974955 }, { "content_chunks": 30, "controlled_output": null, "decode_tok_s": 147.8612019927061, "e2e_ms": 617.789277996053, "effective_prefill_tok_s": 23927.52755754805, "generation_ms": 493.7062530007097, "mean_inter_token_latency_ms": 6.763099356174106, "mean_time_per_output_token_ms": 6.763099356174106, "output_token_source": "usage", "output_tokens": 74, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 28, "time_to_first_output_ms": 124.0830249953433, "tpot_ms": 6.763099356174106, "ttft_ms": 124.08365499868523, "visible_content_truncated": false, "visible_generation_ms": 493.70562299736775 }, { "content_chunks": 32, "controlled_output": null, "decode_tok_s": 138.24966358177087, "e2e_ms": 647.5884280007449, "effective_prefill_tok_s": 24833.091764000757, "generation_ms": 528.0302180035505, "mean_inter_token_latency_ms": 7.233290657582884, "mean_time_per_output_token_ms": 7.233290657582884, "output_token_source": "usage", "output_tokens": 74, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 29, "time_to_first_output_ms": 119.55820999719435, "tpot_ms": 7.233290657582884, "ttft_ms": 119.55851999664446, "visible_content_truncated": false, "visible_generation_ms": 528.0299080041004 }, { "content_chunks": 15, "controlled_output": null, "decode_tok_s": 139.74000254549946, "e2e_ms": 357.46956700313604, "effective_prefill_tok_s": 24473.131962676533, "generation_ms": 236.15285100095207, "mean_inter_token_latency_ms": 7.1561470000288505, "mean_time_per_output_token_ms": 7.1561470000288505, "output_token_source": "usage", "output_tokens": 34, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 30, "time_to_first_output_ms": 121.31671600218397, "tpot_ms": 7.1561470000288505, "ttft_ms": 121.31729600514518, "visible_content_truncated": false, "visible_generation_ms": 236.15227099799085 }, { "content_chunks": 17, "controlled_output": null, "decode_tok_s": 127.07019494203837, "e2e_ms": 398.18144299351843, "effective_prefill_tok_s": 24188.726292886564, "generation_ms": 275.4383119972772, "mean_inter_token_latency_ms": 7.8696660570650625, "mean_time_per_output_token_ms": 7.8696660570650625, "output_token_source": "usage", "output_tokens": 36, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 31, "time_to_first_output_ms": 122.74313099624123, "tpot_ms": 7.8696660570650625, "ttft_ms": 122.74338099814486, "visible_content_truncated": false, "visible_generation_ms": 275.4380619953736 }, { "content_chunks": 21, "controlled_output": null, "decode_tok_s": 143.31560385338523, "e2e_ms": 474.2346980056027, "effective_prefill_tok_s": 25080.943485420292, "generation_ms": 355.85797100065975, "mean_inter_token_latency_ms": 6.97760727452274, "mean_time_per_output_token_ms": 6.97760727452274, "output_token_source": "usage", "output_tokens": 52, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 32, "time_to_first_output_ms": 118.37672700494295, "tpot_ms": 6.97760727452274, "ttft_ms": 118.37697699957062, "visible_content_truncated": false, "visible_generation_ms": 355.8577210060321 }, { "content_chunks": 18, "controlled_output": null, "decode_tok_s": 114.93274567667875, "e2e_ms": 428.45063399727223, "effective_prefill_tok_s": 23958.094234756423, "generation_ms": 304.52591899666004, "mean_inter_token_latency_ms": 8.700740542761716, "mean_time_per_output_token_ms": 8.700740542761716, "output_token_source": "usage", "output_tokens": 36, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 33, "time_to_first_output_ms": 123.9247150006122, "tpot_ms": 8.700740542761716, "ttft_ms": 123.92501499562059, "visible_content_truncated": false, "visible_generation_ms": 304.52561900165165 }, { "content_chunks": 26, "controlled_output": null, "decode_tok_s": 137.78556513719002, "e2e_ms": 552.0714680023957, "effective_prefill_tok_s": 23968.703154295, "generation_ms": 428.2016040015151, "mean_inter_token_latency_ms": 7.257654305110425, "mean_time_per_output_token_ms": 7.257654305110425, "output_token_source": "usage", "output_tokens": 60, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 34, "time_to_first_output_ms": 123.86986400088063, "tpot_ms": 7.257654305110425, "ttft_ms": 123.87023400515318, "visible_content_truncated": false, "visible_generation_ms": 428.2012339972425 }, { "content_chunks": 5, "controlled_output": null, "decode_tok_s": 90.68025791161197, "e2e_ms": 190.1020459990832, "effective_prefill_tok_s": 23956.010534759356, "generation_ms": 66.16655199468369, "mean_inter_token_latency_ms": 11.027758665780615, "mean_time_per_output_token_ms": 11.027758665780615, "output_token_source": "usage", "output_tokens": 7, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 35, "time_to_first_output_ms": 123.93549400439952, "tpot_ms": 11.027758665780615, "ttft_ms": 123.93601400253829, "visible_content_truncated": false, "visible_generation_ms": 66.16603199654492 }, { "content_chunks": 32, "controlled_output": null, "decode_tok_s": 157.26724939223175, "e2e_ms": 649.1538139962358, "effective_prefill_tok_s": 24458.40430785881, "generation_ms": 527.7640470012557, "mean_inter_token_latency_ms": 6.3586029759187435, "mean_time_per_output_token_ms": 6.3586029759187435, "output_token_source": "usage", "output_tokens": 84, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 36, "time_to_first_output_ms": 121.38976699498016, "tpot_ms": 6.3586029759187435, "ttft_ms": 121.39022699557245, "visible_content_truncated": false, "visible_generation_ms": 527.7635870006634 }, { "content_chunks": 3, "controlled_output": null, "decode_tok_s": 81.29469278202373, "e2e_ms": 176.82276199775515, "effective_prefill_tok_s": 23264.550345863776, "generation_ms": 49.203703994862735, "mean_inter_token_latency_ms": 12.300925998715684, "mean_time_per_output_token_ms": 12.300925998715684, "output_token_source": "usage", "output_tokens": 5, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 37, "time_to_first_output_ms": 127.61905800289242, "tpot_ms": 12.300925998715684, "ttft_ms": 127.61958799819695, "visible_content_truncated": false, "visible_generation_ms": 49.2031739995582 }, { "content_chunks": 28, "controlled_output": null, "decode_tok_s": 134.52989745468852, "e2e_ms": 605.0358259963105, "effective_prefill_tok_s": 22961.19033778935, "generation_ms": 475.73068299971055, "mean_inter_token_latency_ms": 7.433291921870477, "mean_time_per_output_token_ms": 7.433291921870477, "output_token_source": "usage", "output_tokens": 65, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 38, "time_to_first_output_ms": 129.3051429965999, "tpot_ms": 7.433291921870477, "ttft_ms": 129.3055229980382, "visible_content_truncated": false, "visible_generation_ms": 475.73030299827224 }, { "content_chunks": 21, "controlled_output": null, "decode_tok_s": 138.93303502670918, "e2e_ms": 464.15260399953695, "effective_prefill_tok_s": 23589.67829067541, "generation_ms": 338.29247299581766, "mean_inter_token_latency_ms": 7.197712191400376, "mean_time_per_output_token_ms": 7.197712191400376, "output_token_source": "usage", "output_tokens": 48, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 39, "time_to_first_output_ms": 125.86013100371929, "tpot_ms": 7.197712191400376, "ttft_ms": 125.86050100071589, "visible_content_truncated": false, "visible_generation_ms": 338.29210299882106 }, { "content_chunks": 35, "controlled_output": null, "decode_tok_s": 148.3786984836116, "e2e_ms": 704.2759780015331, "effective_prefill_tok_s": 23813.351385109258, "generation_ms": 579.5980210023117, "mean_inter_token_latency_ms": 6.739511872119904, "mean_time_per_output_token_ms": 6.739511872119904, "output_token_source": "usage", "output_tokens": 87, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 40, "time_to_first_output_ms": 124.67795699922135, "tpot_ms": 6.739511872119904, "ttft_ms": 124.67842700425535, "visible_content_truncated": false, "visible_generation_ms": 579.5975509972777 }, { "content_chunks": 15, "controlled_output": null, "decode_tok_s": 141.70933062358117, "e2e_ms": 360.3304969947203, "effective_prefill_tok_s": 24658.9046023218, "generation_ms": 239.92774399812333, "mean_inter_token_latency_ms": 7.056698352885981, "mean_time_per_output_token_ms": 7.056698352885981, "output_token_source": "usage", "output_tokens": 35, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 41, "time_to_first_output_ms": 120.40275299659697, "tpot_ms": 7.056698352885981, "ttft_ms": 120.40307299321285, "visible_content_truncated": false, "visible_generation_ms": 239.92742400150746 }, { "content_chunks": 4, "controlled_output": null, "decode_tok_s": 60.9909411708306, "e2e_ms": 184.60641799902078, "effective_prefill_tok_s": 24944.777856949324, "generation_ms": 65.58350999694085, "mean_inter_token_latency_ms": 16.395877499235212, "mean_time_per_output_token_ms": 16.395877499235212, "output_token_source": "usage", "output_tokens": 5, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 42, "time_to_first_output_ms": 119.02290800207993, "tpot_ms": 16.395877499235212, "ttft_ms": 119.02319799992256, "visible_content_truncated": false, "visible_generation_ms": 65.58321999909822 }, { "content_chunks": 4, "controlled_output": null, "decode_tok_s": 81.0419301672689, "e2e_ms": 170.95820300164632, "effective_prefill_tok_s": 24415.910202242972, "generation_ms": 49.35716599720763, "mean_inter_token_latency_ms": 12.339291499301908, "mean_time_per_output_token_ms": 12.339291499301908, "output_token_source": "usage", "output_tokens": 5, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 43, "time_to_first_output_ms": 121.6010370044387, "tpot_ms": 12.339291499301908, "ttft_ms": 121.6013470038888, "visible_content_truncated": false, "visible_generation_ms": 49.35685599775752 }, { "content_chunks": 4, "controlled_output": null, "decode_tok_s": 60.96958715176616, "e2e_ms": 192.51080499816453, "effective_prefill_tok_s": 23395.57773158647, "generation_ms": 65.60647999867797, "mean_inter_token_latency_ms": 16.401619999669492, "mean_time_per_output_token_ms": 16.401619999669492, "output_token_source": "usage", "output_tokens": 5, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 44, "time_to_first_output_ms": 126.90432499948656, "tpot_ms": 16.401619999669492, "ttft_ms": 126.90474499686388, "visible_content_truncated": false, "visible_generation_ms": 65.60606000130065 }, { "content_chunks": 115, "controlled_output": null, "decode_tok_s": 132.51468339459362, "e2e_ms": 2047.0468740022625, "effective_prefill_tok_s": 24190.95533497949, "generation_ms": 1924.3150530019193, "mean_inter_token_latency_ms": 7.546333541183997, "mean_time_per_output_token_ms": 7.546333541183997, "output_token_source": "usage", "output_tokens": 256, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 45, "time_to_first_output_ms": 122.73182100034319, "tpot_ms": 7.546333541183997, "ttft_ms": 122.73235100292368, "visible_content_truncated": false, "visible_generation_ms": 1924.3145229993388 }, { "content_chunks": 31, "controlled_output": null, "decode_tok_s": 163.80023588618587, "e2e_ms": 630.4124010057421, "effective_prefill_tok_s": 24002.077840383055, "generation_ms": 506.7147770023439, "mean_inter_token_latency_ms": 6.104997313281252, "mean_time_per_output_token_ms": 6.104997313281252, "output_token_source": "usage", "output_tokens": 84, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 46, "time_to_first_output_ms": 123.69762400339823, "tpot_ms": 6.104997313281252, "ttft_ms": 123.69803400360979, "visible_content_truncated": false, "visible_generation_ms": 506.71436700213235 }, { "content_chunks": 14, "controlled_output": null, "decode_tok_s": 129.59767354360238, "e2e_ms": 347.34570299769985, "effective_prefill_tok_s": 24025.651400278788, "generation_ms": 223.76944899588125, "mean_inter_token_latency_ms": 7.716187896409698, "mean_time_per_output_token_ms": 7.716187896409698, "output_token_source": "usage", "output_tokens": 30, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 47, "time_to_first_output_ms": 123.5762540018186, "tpot_ms": 7.716187896409698, "ttft_ms": 123.57681400317233, "visible_content_truncated": false, "visible_generation_ms": 223.76888899452752 }, { "content_chunks": 19, "controlled_output": null, "decode_tok_s": 122.54347139890305, "e2e_ms": 432.8391100061708, "effective_prefill_tok_s": 24188.343986661548, "generation_ms": 310.094039006799, "mean_inter_token_latency_ms": 8.160369447547343, "mean_time_per_output_token_ms": 8.160369447547343, "output_token_source": "usage", "output_tokens": 39, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 48, "time_to_first_output_ms": 122.74507099937182, "tpot_ms": 8.160369447547343, "ttft_ms": 122.74538100609789, "visible_content_truncated": false, "visible_generation_ms": 310.0937290000729 }, { "content_chunks": 22, "controlled_output": null, "decode_tok_s": 159.79757822804666, "e2e_ms": 493.50169200624805, "effective_prefill_tok_s": 25155.33248260817, "generation_ms": 375.4750270018121, "mean_inter_token_latency_ms": 6.257917116696869, "mean_time_per_output_token_ms": 6.257917116696869, "output_token_source": "usage", "output_tokens": 61, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 49, "time_to_first_output_ms": 118.02666500443593, "tpot_ms": 6.257917116696869, "ttft_ms": 118.02710500342073, "visible_content_truncated": false, "visible_generation_ms": 375.4745870028273 }, { "content_chunks": 89, "controlled_output": null, "decode_tok_s": 148.61301959071818, "e2e_ms": 1614.1991350013996, "effective_prefill_tok_s": 24662.224913342092, "generation_ms": 1493.8125920016319, "mean_inter_token_latency_ms": 6.7288855495569, "mean_time_per_output_token_ms": 6.7288855495569, "output_token_source": "usage", "output_tokens": 223, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 50, "time_to_first_output_ms": 120.3865429997677, "tpot_ms": 6.7288855495569, "ttft_ms": 120.38715300150216, "visible_content_truncated": false, "visible_generation_ms": 1493.8119819998974 }, { "content_chunks": 47, "controlled_output": null, "decode_tok_s": 123.96355921226757, "e2e_ms": 901.2732580013108, "effective_prefill_tok_s": 24994.68831249027, "generation_ms": 782.4880199987092, "mean_inter_token_latency_ms": 8.066886804110403, "mean_time_per_output_token_ms": 8.066886804110403, "output_token_source": "usage", "output_tokens": 98, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 51, "time_to_first_output_ms": 118.78523800260155, "tpot_ms": 8.066886804110403, "ttft_ms": 118.78576799790608, "visible_content_truncated": false, "visible_generation_ms": 782.4874900034047 }, { "content_chunks": 56, "controlled_output": null, "decode_tok_s": 162.20147798924086, "e2e_ms": 1078.6749920007423, "effective_prefill_tok_s": 24123.826914013433, "generation_ms": 955.6016500064288, "mean_inter_token_latency_ms": 6.165171935525347, "mean_time_per_output_token_ms": 6.165171935525347, "output_token_source": "usage", "output_tokens": 156, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 52, "time_to_first_output_ms": 123.07334199431352, "tpot_ms": 6.165171935525347, "ttft_ms": 123.07374199735932, "visible_content_truncated": false, "visible_generation_ms": 955.601250003383 }, { "content_chunks": 3, "controlled_output": null, "decode_tok_s": 81.25979498229087, "e2e_ms": 169.48487799527356, "effective_prefill_tok_s": 24688.16679264812, "generation_ms": 49.22483499831287, "mean_inter_token_latency_ms": 12.306208749578218, "mean_time_per_output_token_ms": 12.306208749578218, "output_token_source": "usage", "output_tokens": 5, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 53, "time_to_first_output_ms": 120.26004299696069, "tpot_ms": 12.306208749578218, "ttft_ms": 120.26056299509946, "visible_content_truncated": false, "visible_generation_ms": 49.2243150001741 }, { "content_chunks": 16, "controlled_output": null, "decode_tok_s": 147.08325733563373, "e2e_ms": 380.85647499974584, "effective_prefill_tok_s": 24236.853207700362, "generation_ms": 258.3570740025607, "mean_inter_token_latency_ms": 6.79887036848844, "mean_time_per_output_token_ms": 6.79887036848844, "output_token_source": "usage", "output_tokens": 39, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 54, "time_to_first_output_ms": 122.49940099718515, "tpot_ms": 6.79887036848844, "ttft_ms": 122.49972100107698, "visible_content_truncated": false, "visible_generation_ms": 258.35675399866886 }, { "content_chunks": 33, "controlled_output": null, "decode_tok_s": 144.3793154428571, "e2e_ms": 666.8558429955738, "effective_prefill_tok_s": 24806.566640876707, "generation_ms": 547.1697920002043, "mean_inter_token_latency_ms": 6.926199898736764, "mean_time_per_output_token_ms": 6.926199898736764, "output_token_source": "usage", "output_tokens": 80, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 55, "time_to_first_output_ms": 119.68605099536944, "tpot_ms": 6.926199898736764, "ttft_ms": 119.68672100192634, "visible_content_truncated": false, "visible_generation_ms": 547.1691219936474 }, { "content_chunks": 46, "controlled_output": null, "decode_tok_s": 164.15095556468617, "e2e_ms": 887.5807119984529, "effective_prefill_tok_s": 24742.787926755987, "generation_ms": 767.5861499956227, "mean_inter_token_latency_ms": 6.091953571393831, "mean_time_per_output_token_ms": 6.091953571393831, "output_token_source": "usage", "output_tokens": 127, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 56, "time_to_first_output_ms": 119.99456200283021, "tpot_ms": 6.091953571393831, "ttft_ms": 119.99499200464925, "visible_content_truncated": false, "visible_generation_ms": 767.5857199938037 }, { "content_chunks": 42, "controlled_output": null, "decode_tok_s": 136.11340775363055, "e2e_ms": 816.8407760022092, "effective_prefill_tok_s": 24971.956173786806, "generation_ms": 697.9474070030847, "mean_inter_token_latency_ms": 7.346814810558786, "mean_time_per_output_token_ms": 7.346814810558786, "output_token_source": "usage", "output_tokens": 96, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 57, "time_to_first_output_ms": 118.8933689991245, "tpot_ms": 7.346814810558786, "ttft_ms": 118.89378900377778, "visible_content_truncated": false, "visible_generation_ms": 697.9469869984314 }, { "content_chunks": 10, "controlled_output": null, "decode_tok_s": 111.11291070182901, "e2e_ms": 272.34899200266227, "effective_prefill_tok_s": 24876.10751695148, "generation_ms": 152.99752200371586, "mean_inter_token_latency_ms": 8.999854235512698, "mean_time_per_output_token_ms": 8.999854235512698, "output_token_source": "usage", "output_tokens": 18, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 58, "time_to_first_output_ms": 119.35146999894641, "tpot_ms": 8.999854235512698, "ttft_ms": 119.35181000444572, "visible_content_truncated": false, "visible_generation_ms": 152.99718199821655 }, { "content_chunks": 27, "controlled_output": null, "decode_tok_s": 146.38850757345398, "e2e_ms": 565.7820149936015, "effective_prefill_tok_s": 24384.417075304973, "generation_ms": 444.0239269970334, "mean_inter_token_latency_ms": 6.831137338415899, "mean_time_per_output_token_ms": 6.831137338415899, "output_token_source": "usage", "output_tokens": 66, "planned_context_tokens": 4096, "prompt_tokens": 2969, "reasoning_chunks": 0, "request_canary": null, "request_index": 59, "time_to_first_output_ms": 121.75808799656807, "tpot_ms": 6.831137338415899, "ttft_ms": 121.7584479963989, "visible_content_truncated": false, "visible_generation_ms": 444.0235669972026 } ], "requests": 60, "response_words": 0, "responses_received": 60, "run_id": "benchmark-20260908T052955Z", "schema": "anvil-serving.benchmark/v1", "serve_flags": { "no_thinking": false, "reasoning_effort": null, "shared_prefix_burst": false, "thinking_mode": "disabled" }, "shared_prefix_tokens": 8000, "source_recipe": null, "started_at": "2026-09-08T05:29:55Z", "started_at_unix_ns": 1788845395017273152, "tier": null, "timestamp_precision": "nanosecond-unix-fields; ISO labels are second precision", "timing_methodology": { "artifact_timestamps": "time.time_ns Unix timestamps; independent artifacts may align them only when the operator supplied the same explicit clock_domain_id", "clock": "client time.perf_counter", "controlled_output": "--response-words adherence is recorded; observe preserves historical behavior while strict rejects non-exact or truncated completions; visible response capture is capped at 8192 characters", "decode": "usage completion tokens after the first token divided by client-observed generation time; completion tokens may include reasoning tokens", "distributions": "request-level mean, sample standard deviation, normal-approximation 95% confidence interval for the mean, min, p25, p50, p75, p90, p95, p99, and max using the nearest-rank percentile method", "effective_prefill": "usage.prompt_tokens divided by client-observed time to first output; includes queueing, scheduling, prefill, and first-token work", "generation": "client-observed E2E minus time to first output", "mean_inter_token_latency": "legacy key for a TPOT proxy: client-observed generation time divided by completion tokens after the first token; not raw per-token ITL", "p99": "numeric nearest-rank p99 is always retained for compatibility; it is descriptive only below 100 eligible samples", "request_canaries": "when enabled, each response must begin with its own synthetic marker and contain no other ANVIL_REQ marker; bounded/truncated captures fail closed and are excluded from performance metrics", "time_to_first_output": "request start through the first non-empty streamed reasoning or content delta", "tpot": "client-observed mean time per output token after the first output; identical to the legacy mean_inter_token_latency key", "ttft": "request start through the first non-empty streamed content delta" }, "wall_clock_ms": 38018.48291900387 }