Qwen3.5-35B-A3B benchmark on an M4 Max · 36 GB

{
  "runId": "run_95732305-b7df-4c1e-8d82-0fa1ef5420eb",
  "bundleId": "mlx-qwen3.5-35b-a3b-4bit-d6c349",
  "status": "verified",
  "promptTokens": 40960,
  "completionTokens": 10240,
  "contextLength": 5120,
  "harness": {
    "version": "0.1.20",
    "gitSha": "unknown"
  },
  "runtime": {
    "name": "mlx_lm",
    "version": "0.31.2",
    "buildFlags": null
  },
  "model": {
    "displayName": "Qwen3.5-35B-A3B",
    "format": "mlx",
    "quant": "4bit",
    "architecture": "qwen3_5_moe",
    "source": "NexVeridian/Qwen3.5-35B-A3B-4bit",
    "fileSizeBytes": 19509025811,
    "lab": {
      "name": "Qwen",
      "slug": "qwen"
    },
    "quantizedBy": {
      "name": "Elijah McMorris",
      "slug": "nexveridian"
    }
  },
  "device": {
    "cpu": "Apple M4 Max",
    "cpuCores": 14,
    "gpu": "Apple M4 Max",
    "gpuCores": 32,
    "gpuCount": 1,
    "ramGb": 36,
    "osName": "macOS",
    "osVersion": "15.6.1"
  },
  "decodeTpsMean": 104,
  "prefillTpsMean": 1461.7,
  "ttftP50Ms": 2801.88,
  "idleTpsMean": 537,
  "peakRssMb": 21504,
  "trialsPassed": 10,
  "trialsTotal": 10,
  "runnabilityScore": 0.6730849999999999,
  "bundleSha256": "9d12c8ec97aca46fbd9821c945fce00a0a10e44a9c1fa8a97aa8e9de811edfe8",
  "createdAt": "2026-04-11T03:16:33.052Z"
}