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

{
  "runId": "run_85b57f64-d5c0-4dc5-a73b-b3952cc1d611",
  "bundleId": "mlx-qwen3.8-27b-4bit-895d07",
  "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": "Qwen3.8-27B",
    "format": "mlx",
    "quant": "4bit",
    "architecture": "qwen3_5",
    "source": "mlx-community/Qwen3.8-27B-4bit",
    "fileSizeBytes": 16054541349,
    "lab": {
      "name": "Qwen",
      "slug": "qwen"
    },
    "quantizedBy": {
      "name": "MLX Community",
      "slug": "mlx-community"
    }
  },
  "device": {
    "cpu": "Apple M4 Max",
    "cpuCores": 14,
    "gpu": "Apple M4 Max",
    "gpuCores": 32,
    "gpuCount": 1,
    "ramGb": 36,
    "osName": "macOS",
    "osVersion": "27.0"
  },
  "decodeTpsMean": 19.9,
  "prefillTpsMean": 197.6,
  "ttftP50Ms": 20725.6,
  "idleTpsMean": 3840,
  "peakRssMb": 18432,
  "trialsPassed": 10,
  "trialsTotal": 10,
  "runnabilityScore": 0.4042142857142857,
  "bundleSha256": "c8fe07f7a86f0bbb9e41f6c6a2726fc35b26fc989a0f7ef7d540b0cc6f9519a8",
  "createdAt": "2026-08-14T20:20:10.750Z"
}