{
  "base_url": "http://127.0.0.1:39053/v1",
  "completed": 4,
  "concurrency": 1,
  "context_distribution": {
    "131072": 4
  },
  "context_seed": 0,
  "context_tokens": 131072,
  "engine": "vllm",
  "failed": 0,
  "failures": [],
  "finished_at": "2026-08-01T12:07:29Z",
  "gpu": "dual-rtx-pro-6000-blackwell-max-q-tp2",
  "manifest": null,
  "max_context_tokens": 262144,
  "max_tokens": 128,
  "measurement_protocol": "capacity-v3",
  "metrics": {
    "content_chunks": 222,
    "content_chunks_s": null,
    "decode_tok_s_p50": 64.96274329680224,
    "decode_tok_s_p95": 66.19334989240683,
    "e2e_p50_ms": 15371.504900000218,
    "e2e_p95_ms": 15751.761199999237,
    "effective_prefill_tok_s_p50": 8570.368793121388,
    "effective_prefill_tok_s_p95": 8618.176684232738,
    "generation_p50_ms": 778.3833000012237,
    "generation_p95_ms": 1078.9484999986598,
    "mean_inter_token_latency_ms_p50": 15.262417647082817,
    "mean_inter_token_latency_ms_p95": 15.413549999980855,
    "output_token_sources": {
      "usage": 4
    },
    "output_tokens": 227,
    "prefix_cache_hit_avg": null,
    "prompt_token_samples": 4,
    "prompt_tokens": 501783,
    "prompt_tokens_p50": 125444.0,
    "prompt_tokens_p95": 125450.0,
    "throughput_tok_s": 3.6681812376924667,
    "ttft_p50_ms": 14593.121599998994,
    "ttft_p95_ms": 14672.812700000577,
    "ttft_samples": 4
  },
  "model": "qwen35-122b-a10b-nvfp4-tp2",
  "request_timings": [
    {
      "decode_tok_s": 66.19334989240683,
      "e2e_ms": 15235.341600000538,
      "effective_prefill_tok_s": 8618.176684232738,
      "generation_ms": 679.8266000005242,
      "mean_inter_token_latency_ms": 15.107257777789426,
      "output_token_source": "usage",
      "output_tokens": 46,
      "planned_context_tokens": 131072,
      "prompt_tokens": 125442,
      "request_index": 0,
      "ttft_ms": 14555.515000000014
    },
    {
      "decode_tok_s": 65.52041905308069,
      "e2e_ms": 15371.504900000218,
      "effective_prefill_tok_s": 8596.104619590687,
      "generation_ms": 778.3833000012237,
      "mean_inter_token_latency_ms": 15.262417647082817,
      "output_token_source": "usage",
      "output_tokens": 52,
      "planned_context_tokens": 131072,
      "prompt_tokens": 125444,
      "request_index": 1,
      "ttft_ms": 14593.121599998994
    },
    {
      "decode_tok_s": 64.96274329680224,
      "e2e_ms": 15514.719099999638,
      "effective_prefill_tok_s": 8570.368793121388,
      "generation_ms": 877.4260000009235,
      "mean_inter_token_latency_ms": 15.39343859650743,
      "output_token_source": "usage",
      "output_tokens": 58,
      "planned_context_tokens": 131072,
      "prompt_tokens": 125447,
      "request_index": 2,
      "ttft_ms": 14637.293099998715
    },
    {
      "decode_tok_s": 64.8779807378081,
      "e2e_ms": 15751.761199999237,
      "effective_prefill_tok_s": 8549.826305626804,
      "generation_ms": 1078.9484999986598,
      "mean_inter_token_latency_ms": 15.413549999980855,
      "output_token_source": "usage",
      "output_tokens": 71,
      "planned_context_tokens": 131072,
      "prompt_tokens": 125450,
      "request_index": 3,
      "ttft_ms": 14672.812700000577
    }
  ],
  "requests": 4,
  "run_id": "benchmark-20260801T120627Z",
  "schema": "anvil-serving.benchmark/v1",
  "serve_flags": {
    "no_thinking": false,
    "reasoning_effort": null,
    "shared_prefix_burst": false,
    "thinking_mode": "disabled"
  },
  "source_recipe": null,
  "started_at": "2026-08-01T12:06:27Z",
  "tier": null,
  "timing_methodology": {
    "clock": "client time.perf_counter",
    "decode": "usage completion tokens after the first token divided by client-observed generation time",
    "effective_prefill": "usage.prompt_tokens divided by client-observed TTFT; includes queueing, scheduling, prefill, and first-token work",
    "generation": "client-observed E2E minus TTFT",
    "mean_inter_token_latency": "client-observed generation time divided by completion tokens after the first token; not a raw per-token timestamp trace",
    "ttft": "request start through the first non-empty streamed content delta"
  },
  "wall_clock_ms": 61883.529000000635
}