{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreihjxrxrusawcrwafooqunphod5p253pmp37g6fgow2e5qrrs636au",
    "uri": "at://did:plc:pgryn3ephfd2xgft23qokfzt/app.bsky.feed.post/3mpiuhy4ls5j2"
  },
  "path": "/t/we-all-start-somewhere/177233#post_7",
  "publishedAt": "2026-06-30T11:07:09.000Z",
  "site": "https://discuss.huggingface.co",
  "textContent": "Your setup is well-suited for local AI. An RTX 5070 with 12GB VRAM can comfortably run many 7B–14B quantized models, making it a great option for working offline while traveling. I’d keep a capable local model for unreliable connections and use cloud models only when you need larger models or higher-quality reasoning. It’s a practical hybrid setup that gives you flexibility wherever you",
  "title": "We all start somewhere"
}