{
  "date": "2026-10-22",
  "scope": "Same Flash-Next IQ3_S model and speed protocol, GPU selection only; no new accuracy evaluation.",
  "originalArchiveSha256": "1a0ef98f0007de143ee63cde094ef376059a5b09379c2ff2fe818c85c14cf65e",
  "originalDual": [
    {
      "inputTokens": 84,
      "generationTokensPerSecond": 48.0,
      "generationRange": [
        34.7,
        76.4
      ],
      "serverPromptSeconds": 1.785,
      "firstContentSeconds": 1.8052198940131348,
      "replySeconds": 4.369044443999883,
      "replyRangeSeconds": [
        3.4626010450010654,
        7.83752790701692
      ],
      "samples": 3
    },
    {
      "inputTokens": 6586,
      "generationTokensPerSecond": 70.4,
      "generationRange": [
        69.7,
        80.6
      ],
      "serverPromptSeconds": 12.9026,
      "firstContentSeconds": 12.934985511004925,
      "replySeconds": 14.579910433996702,
      "replyRangeSeconds": [
        14.495574722997844,
        23.170635463000508
      ],
      "samples": 3
    },
    {
      "inputTokens": 26087,
      "generationTokensPerSecond": 71.0,
      "generationRange": [
        70.9,
        72.7
      ],
      "serverPromptSeconds": 31.2185,
      "firstContentSeconds": 31.31999467301648,
      "replySeconds": 33.09755272499751,
      "replyRangeSeconds": [
        32.990750141005265,
        33.11324697799864
      ],
      "samples": 3
    }
  ],
  "single": [
    {
      "inputTokens": 84,
      "generationTokensPerSecond": 43.6,
      "generationRange": [
        25.4,
        58.1
      ],
      "serverPromptSeconds": 0.8148,
      "firstContentSeconds": 0.8437469839700498,
      "replySeconds": 3.7535743420012295,
      "replyRangeSeconds": [
        2.9644604650093243,
        6.923110231000464
      ],
      "samples": 3
    },
    {
      "inputTokens": 6586,
      "generationTokensPerSecond": 64.3,
      "generationRange": [
        60.8,
        69.8
      ],
      "serverPromptSeconds": 6.915,
      "firstContentSeconds": 7.00753857701784,
      "replySeconds": 8.811793345026672,
      "replyRangeSeconds": [
        7.729012405034155,
        20.62433022801997
      ],
      "samples": 3
    },
    {
      "inputTokens": 26087,
      "generationTokensPerSecond": 61.4,
      "generationRange": [
        60.4,
        76.6
      ],
      "serverPromptSeconds": 21.4322,
      "firstContentSeconds": 21.477734987973236,
      "replySeconds": 23.527468657994177,
      "replyRangeSeconds": [
        23.09518915397348,
        23.855831029010005
      ],
      "samples": 3
    }
  ],
  "dualControl": [
    {
      "inputTokens": 84,
      "generationTokensPerSecond": 52.1,
      "generationRange": [
        46.8,
        55.0
      ],
      "serverPromptSeconds": 2.1332,
      "firstContentSeconds": 2.1556457889964804,
      "replySeconds": 4.592119049979374,
      "replyRangeSeconds": [
        4.014223173027858,
        5.330337706953287
      ],
      "samples": 3
    },
    {
      "inputTokens": 6586,
      "generationTokensPerSecond": 74.1,
      "generationRange": [
        70.6,
        75.9
      ],
      "serverPromptSeconds": 13.848,
      "firstContentSeconds": 13.881859596993309,
      "replySeconds": 15.539650093996897,
      "replyRangeSeconds": [
        15.01139640796464,
        19.53027624503011
      ],
      "samples": 3
    },
    {
      "inputTokens": 26087,
      "generationTokensPerSecond": 72.9,
      "generationRange": [
        71.3,
        79.3
      ],
      "serverPromptSeconds": 31.3914,
      "firstContentSeconds": 31.43026108603226,
      "replySeconds": 33.19659616204444,
      "replyRangeSeconds": [
        32.970629458955955,
        33.44298199896002
      ],
      "samples": 3
    }
  ],
  "gpu0": "RTX 5060 Ti 16GB, PCIe 5 x8",
  "gpu1": "RTX 5060 Ti 16GB, PCIe 3 x1, borrowed",
  "onlyInferenceProfileDifference": "gpu: 0 versus [0,1]; engine log destination also differs",
  "requestsIdenticalToOriginal": true,
  "thinking": false,
  "outputTokens": 128,
  "promptCacheTokens": 0,
  "samplesPerCell": 3,
  "rawReasoningRetained": false,
  "speedAnswersRetainedInFollowup": true,
  "accuracyRetested": false,
  "runOrder": [
    "single",
    "dual-control"
  ],
  "productionUnchanged": true,
  "comparisons": [
    {
      "inputTokens": 84,
      "singleGenerationFractionOfFreshDual": 0.836852207293666,
      "freshDualGenerationSpeedup": 1.1949541284403669,
      "singleReplySpeedupVsFreshDual": 1.2233989876249718,
      "singleReplyPercentChangeVsFreshDual": -18.26051761401767,
      "freshDualGenerationChangeVsOriginalPercent": 8.54166666666667
    },
    {
      "inputTokens": 6586,
      "singleGenerationFractionOfFreshDual": 0.8677462887989204,
      "freshDualGenerationSpeedup": 1.1524105754276828,
      "singleReplySpeedupVsFreshDual": 1.763505961333897,
      "singleReplyPercentChangeVsFreshDual": -43.29477631912224,
      "freshDualGenerationChangeVsOriginalPercent": 5.255681818181812
    },
    {
      "inputTokens": 26087,
      "singleGenerationFractionOfFreshDual": 0.8422496570644717,
      "freshDualGenerationSpeedup": 1.1872964169381108,
      "singleReplySpeedupVsFreshDual": 1.410971857814584,
      "singleReplyPercentChangeVsFreshDual": -29.126864262985887,
      "freshDualGenerationChangeVsOriginalPercent": 2.6760563380281877
    }
  ],
  "resources": {
    "single": {
      "observedModelGpus": [
        0
      ],
      "sampledPeakModelVramMiB": {
        "0": 11790,
        "1": 0
      },
      "sampledPeakHostPssGiB": 47.128790855407715,
      "loadSeconds": 7.013078836956993
    },
    "dual-control": {
      "observedModelGpus": [
        0,
        1
      ],
      "sampledPeakModelVramMiB": {
        "0": 11832,
        "1": 13806
      },
      "sampledPeakHostPssGiB": 53.79621696472168,
      "loadSeconds": 31.022083163028583
    }
  }
}
