{
  "publishedAt": "2026-10-30T12:59:45Z",
  "workloads": [
    {
      "id": "short-1028-128",
      "label": "短输入",
      "hardware": "2 台主机 · 8 × RTX PRO 6000 Blackwell 96 GB",
      "inputTokens": 1028,
      "outputTokens": 128
    }
  ],
  "results": [
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "DeepInfer",
      "release": "DP8 / EP8 · 容量 256",
      "concurrency": 1,
      "measuredAt": "2026-10-24T05:26:05.384Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#deepinfer",
      "metrics": {
        "outputTps": 69.27950122456147,
        "ttftSeconds": 0.3818174884384007,
        "requestDecodeTps": 87.09974539540794,
        "e2eSeconds": 1.8471663498135058
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "DeepInfer",
      "release": "DP8 / EP8 · 容量 256",
      "concurrency": 8,
      "measuredAt": "2026-10-24T05:26:17.208Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#deepinfer",
      "metrics": {
        "outputTps": 352.4924290012689,
        "ttftSeconds": 0.5788123290005842,
        "requestDecodeTps": 57.204660767729735,
        "e2eSeconds": 2.8103466238126202
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "DeepInfer",
      "release": "DP8 / EP8 · 容量 256",
      "concurrency": 16,
      "measuredAt": "2026-10-24T05:26:32.105Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#deepinfer",
      "metrics": {
        "outputTps": 516.4540756328256,
        "ttftSeconds": 0.8120113875620518,
        "requestDecodeTps": 44.13954690865559,
        "e2eSeconds": 3.70979863293735
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "DeepInfer",
      "release": "DP8 / EP8 · 容量 256",
      "concurrency": 32,
      "measuredAt": "2026-10-24T05:26:52.369Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#deepinfer",
      "metrics": {
        "outputTps": 713.4327353519229,
        "ttftSeconds": 1.9384263125309644,
        "requestDecodeTps": 37.88055836611406,
        "e2eSeconds": 5.374592396046693
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "DeepInfer",
      "release": "DP8 / EP8 · 容量 256",
      "concurrency": 64,
      "measuredAt": "2026-10-24T05:27:16.054Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#deepinfer",
      "metrics": {
        "outputTps": 1176.656891783985,
        "ttftSeconds": 2.4267916454217584,
        "requestDecodeTps": 31.83063013070199,
        "e2eSeconds": 6.531058352953124
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "DeepInfer",
      "release": "DP8 / EP8 · 容量 256",
      "concurrency": 128,
      "measuredAt": "2026-10-24T05:27:53.129Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#deepinfer",
      "metrics": {
        "outputTps": 1476.8165069710308,
        "ttftSeconds": 3.8219463868908004,
        "requestDecodeTps": 19.675283851379827,
        "e2eSeconds": 10.604164402715014
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "DeepInfer",
      "release": "DP8 / EP8 · 容量 256",
      "concurrency": 256,
      "measuredAt": "2026-10-24T05:28:54.363Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#deepinfer",
      "metrics": {
        "outputTps": 1712.3591962572712,
        "ttftSeconds": 6.382328015316418,
        "requestDecodeTps": 11.276212209912716,
        "e2eSeconds": 18.413425489398485
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "SGLang",
      "release": "TP8 / EP8 · Graph32 · 容量 256",
      "concurrency": 1,
      "measuredAt": "2026-10-24T06:59:12.289Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#sglang",
      "metrics": {
        "outputTps": 108.82130603304233,
        "ttftSeconds": 0.21559697393695387,
        "requestDecodeTps": 132.64050689764582,
        "e2eSeconds": 1.1761901696875157
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "SGLang",
      "release": "TP8 / EP8 · Graph32 · 容量 256",
      "concurrency": 8,
      "measuredAt": "2026-10-24T06:59:23.096Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#sglang",
      "metrics": {
        "outputTps": 397.14898504155383,
        "ttftSeconds": 0.978180208437152,
        "requestDecodeTps": 84.52147679335073,
        "e2eSeconds": 2.4845732017499813
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "SGLang",
      "release": "TP8 / EP8 · Graph32 · 容量 256",
      "concurrency": 16,
      "measuredAt": "2026-10-24T06:59:37.919Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#sglang",
      "metrics": {
        "outputTps": 522.7069605473614,
        "ttftSeconds": 1.5195112247503175,
        "requestDecodeTps": 57.18230834602545,
        "e2eSeconds": 3.760213507312642
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "SGLang",
      "release": "TP8 / EP8 · Graph32 · 容量 256",
      "concurrency": 32,
      "measuredAt": "2026-10-24T07:00:01.146Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#sglang",
      "metrics": {
        "outputTps": 609.8111227024658,
        "ttftSeconds": 2.3637483242499684,
        "requestDecodeTps": 32.089824815346354,
        "e2eSeconds": 6.429801532546718
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "SGLang",
      "release": "TP8 / EP8 · Graph32 · 容量 256",
      "concurrency": 64,
      "measuredAt": "2026-10-24T07:00:46.671Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#sglang",
      "metrics": {
        "outputTps": 576.3740347273442,
        "ttftSeconds": 3.846484140077962,
        "requestDecodeTps": 13.739376921137767,
        "e2eSeconds": 13.455711105765431
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "SGLang",
      "release": "TP8 / EP8 · Graph32 · 容量 256",
      "concurrency": 128,
      "measuredAt": "2026-10-24T07:02:06.160Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#sglang",
      "metrics": {
        "outputTps": 664.9545770295499,
        "ttftSeconds": 6.703715283113311,
        "requestDecodeTps": 8.123945893043244,
        "e2eSeconds": 23.241222698445284
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "SGLang",
      "release": "TP8 / EP8 · Graph32 · 容量 256",
      "concurrency": 256,
      "measuredAt": "2026-10-24T06:58:49.061Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#sglang",
      "metrics": {
        "outputTps": 652.915823058326,
        "ttftSeconds": 12.452743981714832,
        "requestDecodeTps": 3.9317598638548867,
        "e2eSeconds": 46.45364094468175
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "vLLM",
      "release": "TP4 / DP2 / EP8 · 容量 256",
      "concurrency": 1,
      "measuredAt": "2026-10-24T07:26:27.416Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#vllm",
      "metrics": {
        "outputTps": 62.30911319071933,
        "ttftSeconds": 0.2491044809999039,
        "requestDecodeTps": 70.53045411401483,
        "e2eSeconds": 2.054104655500396
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "vLLM",
      "release": "TP4 / DP2 / EP8 · 容量 256",
      "concurrency": 8,
      "measuredAt": "2026-10-24T07:26:40.108Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#vllm",
      "metrics": {
        "outputTps": 283.2238227990696,
        "ttftSeconds": 1.1160730224378312,
        "requestDecodeTps": 56.61954998222409,
        "e2eSeconds": 3.4134676278752067
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "vLLM",
      "release": "TP4 / DP2 / EP8 · 容量 256",
      "concurrency": 16,
      "measuredAt": "2026-10-24T07:26:58.233Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#vllm",
      "metrics": {
        "outputTps": 377.12872343585224,
        "ttftSeconds": 1.7025402223127912,
        "requestDecodeTps": 38.71836398938664,
        "e2eSeconds": 5.053451414469009
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "vLLM",
      "release": "TP4 / DP2 / EP8 · 容量 256",
      "concurrency": 32,
      "measuredAt": "2026-10-24T07:27:24.190Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#vllm",
      "metrics": {
        "outputTps": 519.4016699202651,
        "ttftSeconds": 3.678261087015642,
        "requestDecodeTps": 33.00188412884744,
        "e2eSeconds": 7.587256589499874
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "vLLM",
      "release": "TP4 / DP2 / EP8 · 容量 256",
      "concurrency": 64,
      "measuredAt": "2026-10-24T07:28:02.864Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#vllm",
      "metrics": {
        "outputTps": 679.64826417729,
        "ttftSeconds": 6.739876557179485,
        "requestDecodeTps": 26.948282039271454,
        "e2eSeconds": 11.515537296281197
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "vLLM",
      "release": "TP4 / DP2 / EP8 · 容量 256",
      "concurrency": 128,
      "measuredAt": "2026-10-24T07:29:13.214Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#vllm",
      "metrics": {
        "outputTps": 727.9975655415856,
        "ttftSeconds": 13.660333819097659,
        "requestDecodeTps": 16.809431762128792,
        "e2eSeconds": 21.30732404915237
      }
    },
    {
      "workloadId": "short-1028-128",
      "model": "DeepSeek-V4.1-Flash-1017",
      "engine": "vLLM",
      "release": "TP4 / DP2 / EP8 · 容量 256",
      "concurrency": 256,
      "measuredAt": "2026-10-24T07:25:50.525Z",
      "sourceUrl": "https://github.com/NewbieFoxMe/benchmark/blob/main/docs/datasets/2026-10-24-concurrency256.md#vllm",
      "metrics": {
        "outputTps": 759.5940542481228,
        "ttftSeconds": 21.570092162804748,
        "requestDecodeTps": 7.5839324257096985,
        "e2eSeconds": 40.48934083058599
      }
    }
  ]
}
