LFM2-1.2B-Extract benchmark on a 2× TITAN Xp

{
  "runId": "run_bcf1a31e-4dc7-49f0-994c-d222e23d5378",
  "bundleId": "llamacpp-lfm2-1.2b-extract-f16.gguf-9acf86",
  "status": "verified",
  "promptTokens": 40960,
  "completionTokens": 10240,
  "contextLength": 5120,
  "harness": {
    "version": "0.1.21",
    "gitSha": "4af3357"
  },
  "runtime": {
    "name": "llama.cpp",
    "version": "b1",
    "buildFlags": "metal"
  },
  "model": {
    "displayName": "LFM2-1.2B-Extract",
    "format": "gguf",
    "quant": "f16",
    "architecture": "lfm2",
    "source": "LiquidAI/LFM2-1.2B-Extract-GGUF:LFM2-1.2B-Extract-F16.gguf",
    "fileSizeBytes": 2343325408,
    "lab": {
      "name": "Liquid AI",
      "slug": "liquid"
    },
    "quantizedBy": {
      "name": "Liquid AI",
      "slug": "liquid"
    }
  },
  "device": {
    "cpu": "Intel(R) Core(TM) i7-5930K CPU @ 3.50GHz",
    "cpuCores": 12,
    "gpu": "NVIDIA TITAN Xp",
    "gpuCores": 0,
    "gpuCount": 2,
    "ramGb": 31,
    "osName": "Ubuntu 24.04.4 LTS",
    "osVersion": "24.04"
  },
  "decodeTpsMean": 156.2,
  "prefillTpsMean": 4109.3,
  "ttftP50Ms": 996.79,
  "idleTpsMean": 709.6,
  "peakRssMb": 2583.4,
  "trialsPassed": 10,
  "trialsTotal": 10,
  "runnabilityScore": 0.9651218677995391,
  "bundleSha256": "b0b8e893ce28d74c6064ed9e85de1cff2e235f0400956c1defa8e4b007f2f00b",
  "createdAt": "2026-04-14T02:08:54.418Z"
}