{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreiams27a2opb2b7vlsahiflxopbbk26tcghmhddnk4essi7j3vnedu",
    "uri": "at://did:plc:7fzsvgbjbyp2vxleodbcwyif/app.bsky.feed.post/3mfd6ufn34ee2"
  },
  "path": "/2026/02/17/gguf-runner-blogpost.html",
  "publishedAt": "2026-02-17T00:00:00.000Z",
  "site": "https://jens.dev",
  "tags": [
    "https://github.com/apimeister/gguf-runner"
  ],
  "textContent": "I’ve been playing with local LLMs again.\n\nNot in the “let’s build a platform” way. More in the “I want a tiny tool I can keep in `~/bin` and forget about” way.\n\nSo I built **gguf-runner** : a small Rust CLI to run **GGUF** models locally, **CPU-only** , with a focus on _low memory overhead_ and a clean “pipes and scripts” workflow.\n\nRepo: https://github.com/apimeister/gguf-runner\n\n## Memory first: GGUF + mmap\n\nThe core idea behind gguf-runner is simple:",
  "title": "gguf-runner: a minimal GGUF CLI"
}