Qwen3-4B-Instruct-2507 benchmark on an M1 Max · 64 GB

{
  "runId": "run_4cb824a6-1891-4408-a277-aac148fe5b5a",
  "bundleId": "llamacpp-qwen3-4b-instruct-2507-q8_0.gguf-07cead",
  "status": "verified",
  "promptTokens": 122880,
  "completionTokens": 30720,
  "contextLength": 5120,
  "harness": {
    "version": "0.1.10",
    "gitSha": "753787f"
  },
  "runtime": {
    "name": "llama.cpp",
    "version": "b8240",
    "buildFlags": "metal"
  },
  "model": {
    "displayName": "Qwen3-4B-Instruct-2507",
    "format": "gguf",
    "quant": "q8_0",
    "architecture": "qwen3",
    "source": "unsloth/Qwen3-4B-Instruct-2507-GGUF:Qwen3-4B-Instruct-2507-Q8_0.gguf",
    "fileSizeBytes": 4280405600,
    "lab": {
      "name": "Qwen",
      "slug": "qwen"
    },
    "quantizedBy": {
      "name": "Unsloth",
      "slug": "unsloth"
    }
  },
  "device": {
    "cpu": "Apple M1 Max",
    "cpuCores": 10,
    "gpu": "Apple M1 Max",
    "gpuCores": 32,
    "gpuCount": 1,
    "ramGb": 64,
    "osName": "macOS",
    "osVersion": "15.6.1"
  },
  "decodeTpsMean": 30.4,
  "prefillTpsMean": 495.6,
  "ttftP50Ms": 8859.4,
  "idleTpsMean": 5134,
  "peakRssMb": 5145,
  "trialsPassed": 30,
  "trialsTotal": 30,
  "runnabilityScore": 0.5965733333333334,
  "bundleSha256": "c02ff91c700e36489605b78944a0cbef56ba561bf1f20a277bb5ac5278055c9c",
  "createdAt": "2026-03-23T21:42:00.684Z"
}