llama.cpp ↗
Run language models efficiently across a wide range of local hardware.
Self-hosted
Get a language model running on your own computer, and start building locally.
The local software is MIT-licensed. You provide hardware and download models under their own licenses. Cloud model services are separate.
Official documentation checked: 2026-10-11. Not a full product test. Terms and availability can change.We rate documented usefulness, ease of starting, and the scope of free access, each out of five. The displayed score is their equal-weight average. These are editorial opinions, not user reviews or hands-on benchmarks.
This is a self-hosted tool. Check operating-system support, model licensing, hardware requirements, and any external API costs before installation.