{
  "evidenceVersion": "local-ollama-speed-v1",
  "capturedAt": "2026-06-01T21:43:25.680Z",
  "hardwareSlug": "rtx-3080-16gb-mobile",
  "modelTag": "qwen3:4b",
  "catalogSlug": "qwen-3-4b",
  "quant": "Q4_K_M",
  "contextSize": 4096,
  "numPredict": 256,
  "prompt": "Write a detailed explanation of how transformer attention works.",
  "scenario": "single-stream",
  "runtime": "ollama-api",
  "command": "POST http://localhost:11434/api/generate model=qwen3:4b temperature=0 top_p=1 seed=42 num_ctx=4096 num_predict=256",
  "env": {
    "os": "Microsoft Windows [Version 10.0.26200.8457]",
    "platform": "win32",
    "arch": "x64",
    "cpu": "AMD Ryzen 9 5900HX with Radeon Graphics        ",
    "cpuThreads": 16,
    "ramGb": 31.9,
    "gpu": {
      "vendor": "nvidia",
      "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
      "driver": "571.96",
      "memoryTotalMb": 16384,
      "memoryUsedMb": 4847,
      "tempC": 64,
      "smClockMhz": 210,
      "powerDrawW": 18.13
    },
    "ollamaVersionCli": "ollama version is 0.24.0",
    "ollamaApiVersion": {
      "version": "0.24.0"
    }
  },
  "ollamaTagDetails": {
    "name": "qwen3:4b",
    "model": "qwen3:4b",
    "modified_at": "2026-06-02T00:43:05.7062155+03:00",
    "size": 2497293931,
    "digest": "359d7dd4bcdab3d86b87d73ac27966f4dbb9f5efdfcc75d34a8764a09474fae7",
    "details": {
      "parent_model": "",
      "format": "gguf",
      "family": "qwen3",
      "families": [
        "qwen3"
      ],
      "parameter_size": "4.0B",
      "quantization_level": "Q4_K_M"
    }
  },
  "notes": null,
  "logHashSha256": "20a38d25dbac9aaa92fd3391cfa2abba7119e0385f99c1ee21266f632a1403d8",
  "runsCaptured": 5,
  "decodeStatsAllRuns": {
    "min": 100.90825311258602,
    "max": 106.4080861499817,
    "median": 103.20566470092126,
    "p5": 101.34216797482874,
    "p95": 105.94514379181226,
    "mean": 103.53864114928463,
    "n": 5
  },
  "decodeStatsSteadyRuns": {
    "min": 100.90825311258602,
    "max": 106.4080861499817,
    "median": 103.64951953002789,
    "p5": 101.2528648508363,
    "p95": 106.06087938135462,
    "mean": 103.65384458065589,
    "n": 4
  },
  "runs": [
    {
      "runIndex": 1,
      "startedAt": "2026-06-01T21:43:07.823Z",
      "finishedAt": "2026-06-01T21:43:13.148Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/qwen3-4b/run1.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/qwen3-4b/run1.log",
      "decodeTokPerSec": 103.0778274237996,
      "prefillTokPerSec": 887.3547511191762,
      "totalMs": 5235.7064,
      "loadMs": 2523.5824,
      "promptEvalMs": 22.5389,
      "evalMs": 2483.5603,
      "promptTokens": 20,
      "responseTokens": 256,
      "responseHashSha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 4847,
        "tempC": 64,
        "smClockMhz": 210,
        "powerDrawW": 18.23
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8140,
        "tempC": 70,
        "smClockMhz": 1485,
        "powerDrawW": 111.17
      }
    },
    {
      "runIndex": 2,
      "startedAt": "2026-06-01T21:43:13.225Z",
      "finishedAt": "2026-06-01T21:43:16.256Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/qwen3-4b/run2.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/qwen3-4b/run2.log",
      "decodeTokPerSec": 100.90825311258602,
      "prefillTokPerSec": 1358.6679619301237,
      "totalMs": 2950.8162,
      "loadMs": 287.6359,
      "promptEvalMs": 14.7203,
      "evalMs": 2536.958,
      "promptTokens": 20,
      "responseTokens": 256,
      "responseHashSha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8140,
        "tempC": 68,
        "smClockMhz": 1485,
        "powerDrawW": 111.17
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8140,
        "tempC": 71,
        "smClockMhz": 1410,
        "powerDrawW": 112.74
      }
    },
    {
      "runIndex": 3,
      "startedAt": "2026-06-01T21:43:16.333Z",
      "finishedAt": "2026-06-01T21:43:19.393Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/qwen3-4b/run3.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/qwen3-4b/run3.log",
      "decodeTokPerSec": 104.09337435913451,
      "prefillTokPerSec": 1503.3298756746194,
      "totalMs": 2961.5744,
      "loadMs": 290.9792,
      "promptEvalMs": 13.3038,
      "evalMs": 2459.3304,
      "promptTokens": 20,
      "responseTokens": 256,
      "responseHashSha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8140,
        "tempC": 71,
        "smClockMhz": 1410,
        "powerDrawW": 106.15
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8140,
        "tempC": 71,
        "smClockMhz": 1440,
        "powerDrawW": 113.79
      }
    },
    {
      "runIndex": 4,
      "startedAt": "2026-06-01T21:43:19.472Z",
      "finishedAt": "2026-06-01T21:43:22.555Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/qwen3-4b/run4.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/qwen3-4b/run4.log",
      "decodeTokPerSec": 106.4080861499817,
      "prefillTokPerSec": 1481.9205690574986,
      "totalMs": 3006.7578,
      "loadMs": 332.2871,
      "promptEvalMs": 13.496,
      "evalMs": 2405.8322,
      "promptTokens": 20,
      "responseTokens": 256,
      "responseHashSha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8140,
        "tempC": 71,
        "smClockMhz": 1755,
        "powerDrawW": 107.37
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8140,
        "tempC": 72,
        "smClockMhz": 1410,
        "powerDrawW": 112.9
      }
    },
    {
      "runIndex": 5,
      "startedAt": "2026-06-01T21:43:22.634Z",
      "finishedAt": "2026-06-01T21:43:25.679Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/qwen3-4b/run5.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/qwen3-4b/run5.log",
      "decodeTokPerSec": 103.20566470092126,
      "prefillTokPerSec": 1460.8776953193478,
      "totalMs": 2956.0613,
      "loadMs": 267.5134,
      "promptEvalMs": 13.6904,
      "evalMs": 2480.484,
      "promptTokens": 20,
      "responseTokens": 256,
      "responseHashSha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8140,
        "tempC": 72,
        "smClockMhz": 1410,
        "powerDrawW": 106.45
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8140,
        "tempC": 73,
        "smClockMhz": 1380,
        "powerDrawW": 113.74
      }
    }
  ],
  "errors": [],
  "uploadEligible": true
}