Bonsai 27B benchmark on an M4 Max · 36 GB

{
  "runId": "run_39d8c039-1021-4fa0-8f6d-7d0898f421e4",
  "bundleId": "llamacpp-bonsai-27b-q1_0.gguf-79271f",
  "status": "verified",
  "promptTokens": 40960,
  "completionTokens": 10240,
  "contextLength": 5120,
  "harness": {
    "version": "0.1.21",
    "gitSha": "48f6137"
  },
  "runtime": {
    "name": "llama.cpp",
    "version": "b9960",
    "buildFlags": "metal"
  },
  "model": {
    "displayName": "Bonsai 27B",
    "format": "gguf",
    "quant": "q1_0",
    "architecture": "qwen35",
    "source": "prism-ml/Bonsai-27B-gguf:Bonsai-27B-Q1_0.gguf",
    "fileSizeBytes": 3803452480,
    "lab": {
      "name": "PrismML",
      "slug": "prismml"
    },
    "quantizedBy": {
      "name": "PrismML",
      "slug": "prismml"
    }
  },
  "device": {
    "cpu": "Apple M4 Max",
    "cpuCores": 14,
    "gpu": "Apple M4 Max",
    "gpuCores": 32,
    "gpuCount": 1,
    "ramGb": 36,
    "osName": "macOS",
    "osVersion": "27.0"
  },
  "decodeTpsMean": 41.5,
  "prefillTpsMean": 205.6,
  "ttftP50Ms": 19900.3,
  "idleTpsMean": 4424.4,
  "peakRssMb": 5175.4,
  "trialsPassed": 10,
  "trialsTotal": 10,
  "runnabilityScore": 0.6530154575892857,
  "bundleSha256": "82fa4dbd00340289b413f6775ecdd1050dc1446adad804c49cd1a2c2e26b508e",
  "createdAt": "2026-07-16T02:55:44.496Z"
}