LFM2.5-VL-450M benchmark on a GeForce RTX 5070 Ti

{
  "runId": "run_ec658984-c993-43ac-aab6-6f646b1bce92",
  "bundleId": "llamacpp-lfm2.5-vl-450m-f16.gguf-b4ef9c",
  "status": "verified",
  "promptTokens": 40960,
  "completionTokens": 10240,
  "contextLength": 5120,
  "harness": {
    "version": "0.1.20",
    "gitSha": "unknown"
  },
  "runtime": {
    "name": "llama.cpp",
    "version": "b1",
    "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": "Intel(R) Core(TM) Ultra 7 265KF",
    "cpuCores": 20,
    "gpu": "NVIDIA GeForce RTX 5070 Ti",
    "gpuCores": 0,
    "gpuCount": 1,
    "ramGb": 23,
    "osName": "Fedora Remix for WSL",
    "osVersion": "Unknown"
  },
  "decodeTpsMean": 548.7,
  "prefillTpsMean": 41262.9,
  "ttftP50Ms": 100.14,
  "idleTpsMean": 1007.6,
  "peakRssMb": 1007.6,
  "trialsPassed": 10,
  "trialsTotal": 10,
  "runnabilityScore": 0.9816648874223601,
  "bundleSha256": "a9cdc0a13f9b15af38ef0a6be1a12f443a0d7e47a073a5d29cc8741acc4efa9d",
  "createdAt": "2026-04-09T06:42:43.966Z"
}