{
  "execution_mode": "live HTTP requests",
  "offered_requests": 6,
  "client_shed_requests": 0,
  "duration_s": 3.428,
  "ttft_slo_ms": 500,
  "ttft_goodput_rps": 0.875,
  "requests": [
    {
      "status": "completed",
      "ttft_ms": 286.882,
      "event_gaps_ms": [
        50.741,
        33.183,
        96.341,
        51.815,
        66.654,
        47.32,
        29.61,
        28.277,
        28.903,
        37.82,
        35.617,
        34.327,
        29.957,
        30.287,
        27.861,
        28.229,
        34.636,
        28.27,
        30.765,
        34.718,
        31.115,
        31.47,
        28.42
      ],
      "e2e_ms": 1163.249,
      "text": "A cache is a temporary storage device that stores frequently accessed data in memory. When a program needs to access a piece of",
      "content_events": 24
    },
    {
      "status": "failed",
      "error": "HTTP Error 429: Too Many Requests",
      "e2e_ms": 1.148
    },
    {
      "status": "failed",
      "error": "HTTP Error 429: Too Many Requests",
      "e2e_ms": 0.775
    },
    {
      "status": "completed",
      "ttft_ms": 42.988,
      "event_gaps_ms": [
        29.627,
        27.375,
        26.108,
        26.265,
        26.75,
        25.804,
        27.251,
        25.603,
        27.745,
        29.644,
        26.108,
        25.73,
        26.063,
        27.207,
        26.141,
        29.924,
        27.141,
        28.426,
        32.937,
        27.496,
        26.547,
        27.139,
        26.988
      ],
      "e2e_ms": 673.037,
      "text": "A cache is a temporary storage device that stores frequently accessed data in memory. When a program needs to access a piece of",
      "content_events": 24
    },
    {
      "status": "failed",
      "error": "HTTP Error 429: Too Many Requests",
      "e2e_ms": 1.011
    },
    {
      "status": "completed",
      "ttft_ms": 41.93,
      "event_gaps_ms": [
        31.872,
        27.103,
        26.454,
        26.67,
        26.131,
        25.989,
        34.774,
        41.762,
        34.019,
        27.688,
        27.436,
        27.602,
        54.374,
        31.225,
        28.678,
        56.612,
        48.569,
        50.115,
        54.832,
        61.134,
        47.739,
        51.438,
        43.673
      ],
      "e2e_ms": 927.88,
      "text": "A cache is a temporary storage device that stores frequently accessed data in memory. When a program needs to access a piece of",
      "content_events": 24
    }
  ],
  "limits": "TTFT-only goodput; SSE events are not necessarily model tokens. Client saturation and failed responses are disclosed.",
  "run_context": {
    "model": "Qwen/Qwen2.5-0.5B-Instruct",
    "revision": "7ae557604adf67be50417f59c2c2f167def9a775",
    "license": "Apache-2.0",
    "device": "M4 Pro / PyTorch MPS",
    "command": "python3 benchmark.py --url http://127.0.0.1:8767/generate --requests 6 --rate 2 --output http-run.json",
    "source_sha256": {
      "benchmark-vllm-sglang-and-tensorrt-llm-on-real-traffic/benchmark.py": "3ecbf214b284bfddf6b72f50987459a719fca3822eb9dd9c29a3fbfe67380785",
      "stream-tokens-over-sse-and-cancel-dropped-requests/server.py": "b019d96d81528e8b44ce25239ee0ba6942505375f8757ca6d6267681353d769a",
      "stream-tokens-over-sse-and-cancel-dropped-requests/live_inference.py": "754d9bf81d04694541e53d5c0472f984c9f046886070c33b376ca071c81c8086"
    },
    "scope": "Six development requests against a single local model slot; not an engine comparison or production load test."
  }
}