A.I.
Llama
Llama is Meta’s open-weight model line for download and self-host. It is separate from the Muse API and Meta AI stack.
Open weight means you can download the model's parameters, the numbers it learned, and run them on your own hardware. Llama is released under Meta's Llama community license, which is not an open-source license like Apache 2.0, so read its terms before commercial use. The current generation is Llama 4: Scout and Maverick, both Mixture of Experts models, meaning only part of the model runs for each token. Meta's newer frontier work ships as the closed Muse family instead, plus the open-weight Muse Glimmer.
Current
| Model | Status | Price | Released |
|---|---|---|---|
| Llama 4Current open-weight generation. Scout and Maverick.Scout-17B-16E · Maverick-17B-128E · Llama 4 Community License | Current, open weights | See page | Not on file |
Prices are US dollars per million tokens (word pieces) in and out unless noted. "Not on file" means this site has no checked value yet.
Prior releases, without full cards here: Llama 3, 3.1, 3.2, 3.3 and earlier. See Meta’s llama-models herd table and Hugging Face meta-llama.