Llama 3.1 8B Instruct benchmark on an M3 Ultra · 256 GB

{
  "runId": "run_f5f019bd-b155-49ac-a713-5938414809c4",
  "bundleId": "mlx-meta-llama-3.1-8b-instruct-4bit-d115d4",
  "status": "verified",
  "promptTokens": 40960,
  "completionTokens": 10240,
  "contextLength": 5120,
  "harness": {
    "version": "0.1.18",
    "gitSha": "unknown"
  },
  "runtime": {
    "name": "mlx_lm",
    "version": "0.31.1",
    "buildFlags": null
  },
  "model": {
    "displayName": "Llama 3.1 8B Instruct",
    "format": "mlx",
    "quant": "4bit",
    "architecture": "llama",
    "source": "mlx-community/Meta-Llama-3.1-8B-Instruct-4bit",
    "fileSizeBytes": 4517488999,
    "lab": {
      "name": "Meta",
      "slug": "meta"
    },
    "quantizedBy": {
      "name": "MLX Community",
      "slug": "mlx-community"
    }
  },
  "device": {
    "cpu": "Apple M3 Ultra",
    "cpuCores": 28,
    "gpu": "Apple M3 Ultra",
    "gpuCores": 60,
    "gpuCount": 1,
    "ramGb": 256,
    "osName": "macOS",
    "osVersion": "26.4"
  },
  "decodeTpsMean": 108.2,
  "prefillTpsMean": 1226.9,
  "ttftP50Ms": 3338.59,
  "idleTpsMean": 2626,
  "peakRssMb": 7182,
  "trialsPassed": 10,
  "trialsTotal": 10,
  "runnabilityScore": 0.8996033618164063,
  "bundleSha256": "baea3119f1bb16a38b53361836fd5fc35074edfa87817ab89f54ed7141f6593f",
  "createdAt": "2026-04-05T00:11:44.592Z"
}