Qwen3.8-27B benchmark on an M4 Max · 36 GB

{
  "runId": "run_e9e14331-5e33-484d-89ff-c11849aee6c1",
  "bundleId": "llamacpp-qwen3.8-27b-q4_k_s.gguf-074a25",
  "status": "verified",
  "promptTokens": 40960,
  "completionTokens": 10240,
  "contextLength": 5120,
  "harness": {
    "version": "0.1.21",
    "gitSha": "unknown"
  },
  "runtime": {
    "name": "llama.cpp",
    "version": "b9960",
    "buildFlags": "metal"
  },
  "model": {
    "displayName": "Qwen3.8-27B",
    "format": "gguf",
    "quant": "q4_k_s",
    "architecture": "qwen35",
    "source": "unsloth/Qwen3.8-27B-GGUF:Qwen3.8-27B-Q4_K_S.gguf",
    "fileSizeBytes": 16121359328,
    "lab": {
      "name": "Qwen",
      "slug": "qwen"
    },
    "quantizedBy": {
      "name": "Unsloth",
      "slug": "unsloth"
    }
  },
  "device": {
    "cpu": "Apple M4 Max",
    "cpuCores": 14,
    "gpu": "Apple M4 Max",
    "gpuCores": 32,
    "gpuCount": 1,
    "ramGb": 36,
    "osName": "macOS",
    "osVersion": "27.0"
  },
  "decodeTpsMean": 18.9,
  "prefillTpsMean": 190.9,
  "ttftP50Ms": 21452.21,
  "idleTpsMean": 1981.4,
  "peakRssMb": 4736,
  "trialsPassed": 10,
  "trialsTotal": 10,
  "runnabilityScore": 0.5527654761904762,
  "bundleSha256": "e0fc79a8a0ec947b2db50b64b3a9b3987b64f8937892f3ffb40416248f295e9a",
  "createdAt": "2026-08-14T19:57:22.398Z"
}