Gemma 4 31B benchmark on an M5 Pro · 64 GB

{
  "runId": "run_a62de28a-2349-4266-b5f5-f35887324276",
  "bundleId": "mlx-gemma-4-31b-8bit-cf1656",
  "status": "verified",
  "promptTokens": 40960,
  "completionTokens": 10240,
  "contextLength": 5120,
  "harness": {
    "version": "0.1.22",
    "gitSha": "ff1d570"
  },
  "runtime": {
    "name": "mlx_lm",
    "version": "0.31.3",
    "buildFlags": null
  },
  "model": {
    "displayName": "Gemma 4 31B",
    "format": "mlx",
    "quant": "8bit",
    "architecture": "gemma4",
    "source": "mlx-community/gemma-4-31b-8bit",
    "fileSizeBytes": 33763120491,
    "lab": {
      "name": "Google",
      "slug": "google"
    },
    "quantizedBy": {
      "name": "MLX Community",
      "slug": "mlx-community"
    }
  },
  "device": {
    "cpu": "Apple M5 Pro",
    "cpuCores": 18,
    "gpu": "Apple M5 Pro",
    "gpuCores": 20,
    "gpuCount": 1,
    "ramGb": 64,
    "osName": "macOS",
    "osVersion": "26.5.1"
  },
  "decodeTpsMean": 7.7,
  "prefillTpsMean": 305.6,
  "ttftP50Ms": 12706.72,
  "idleTpsMean": 32768,
  "peakRssMb": 37888,
  "trialsPassed": 10,
  "trialsTotal": 10,
  "runnabilityScore": 0.2584321428571429,
  "bundleSha256": "3308a101f61f6f98e0b7ffcaeb52945adddb179e79e0b107953d92ab76723361",
  "createdAt": "2026-06-05T16:41:40.494Z"
}