Llama
Open-weight modelsMeta's family of open-weight models — the most widely deployed base for self-hosted AI.
Llama is Meta's family of open-weight large language models. The weights are downloadable, so you can run, fine-tune, and inspect them on your own hardware.
Its scale of adoption is its strongest feature: nearly every serving runtime, quantization tool, and fine-tuning library supports Llama first, so the surrounding ecosystem is the deepest of any open model.
Where it's ideally used
The default open-weight choice when models must run inside your own environment and you want the widest tooling support.
Where it doesn't fit
Its community license carries conditions — review them for large-scale commercial use rather than assuming it is unrestricted.