Qwen3.6-27B benchmark on an M4 Pro · 24 GB

{
  "runId": "run_17bef678-b537-4d22-9b20-5404129df334",
  "bundleId": "mlx-qwen3.6-27b-relaxed-100-mlx-q4-71f14f",
  "status": "verified",
  "promptTokens": 40960,
  "completionTokens": 10240,
  "contextLength": 5120,
  "harness": {
    "version": "0.1.21",
    "gitSha": "unknown"
  },
  "runtime": {
    "name": "mlx_lm",
    "version": "0.31.3",
    "buildFlags": null
  },
  "model": {
    "displayName": "Qwen3.6-27B",
    "format": "mlx",
    "quant": "4bit",
    "architecture": "qwen3_5",
    "source": "s3brr/Qwen3.6-27B-relaxed-100-mlx-q4",
    "fileSizeBytes": 15133043513,
    "lab": {
      "name": "Qwen",
      "slug": "qwen"
    },
    "quantizedBy": null
  },
  "device": {
    "cpu": "Apple M4 Pro",
    "cpuCores": 12,
    "gpu": "Apple M4 Pro",
    "gpuCores": 16,
    "gpuCount": 1,
    "ramGb": 24,
    "osName": "macOS",
    "osVersion": "26.4.1"
  },
  "decodeTpsMean": 15.1,
  "prefillTpsMean": 100.4,
  "ttftP50Ms": 40695.88,
  "idleTpsMean": 3802,
  "peakRssMb": 18432,
  "trialsPassed": 10,
  "trialsTotal": 10,
  "runnabilityScore": 0.251,
  "bundleSha256": "3c29dce9ee9c0cd00cd1ba235681dfa00bd1a7db65035e5bce56ca46bf81ce52",
  "createdAt": "2026-04-26T23:03:53.813Z"
}