Llama
Updated 2026-08-04
Questions this answers
- What is Llama and who makes it?
- I want an open model I can run myself. Where do I start?
- Why is Llama the one everyone builds on?
- Is Llama actually open source?
The Fix
Llama is Meta's family of open-weight models, and the one that did the most to make open models mainstream. Open-weight means you download the model and run it on your own hardware, with no per-token bill and no data leaving your machine.
Llama became the default base that much of the open ecosystem builds on, including fine-tunes like Hermes. It rarely tops the leaderboards outright, but it's capable and everywhere, which makes it a common starting point for building on open weights.
When to Use It
Start here when you want to run a model yourself and want the safest, best-supported open-weight base, with the most tools and community around it. It's a sensible default for local experiments or for fine-tuning to your own use.
Check the license before you ship, since Llama is open-weight rather than fully open source and its terms carry some conditions. If you need the strongest open model, compare it against Qwen, DeepSeek, and Kimi.