LFM2.5-VL-450M benchmark on an M1 Max · 64 GB

{
  "runId": "run_09f1965e-3591-447f-a4c6-e8fafc87ff1e",
  "bundleId": "llamacpp-lfm2.5-vl-450m-f16.gguf-6fb4f1",
  "status": "verified",
  "promptTokens": 40960,
  "completionTokens": 10240,
  "contextLength": 5120,
  "harness": {
    "version": "0.1.20",
    "gitSha": "753787f"
  },
  "runtime": {
    "name": "llama.cpp",
    "version": "b8640",
    "buildFlags": "metal"
  },
  "model": {
    "displayName": "LFM2.5-VL-450M",
    "format": "gguf",
    "quant": "f16",
    "architecture": "lfm2",
    "source": "LiquidAI/LFM2.5-VL-450M-GGUF:LFM2.5-VL-450M-F16.gguf",
    "fileSizeBytes": 711486624,
    "lab": {
      "name": "Liquid AI",
      "slug": "liquid"
    },
    "quantizedBy": {
      "name": "Liquid AI",
      "slug": "liquid"
    }
  },
  "device": {
    "cpu": "Apple M1 Max",
    "cpuCores": 10,
    "gpu": "Apple M1 Max",
    "gpuCores": 32,
    "gpuCount": 1,
    "ramGb": 64,
    "osName": "macOS",
    "osVersion": "15.6.1"
  },
  "decodeTpsMean": 229.8,
  "prefillTpsMean": 10108,
  "ttftP50Ms": 387.39,
  "idleTpsMean": 0.2,
  "peakRssMb": 247,
  "trialsPassed": 10,
  "trialsTotal": 10,
  "runnabilityScore": 0.9983847481863839,
  "bundleSha256": "f1f7cc223c91cde1d79f11e327707374fad5bae037e0eab3efcc0d6290c3ef56",
  "createdAt": "2026-04-09T04:49:39.594Z"
}