Llama 3.2

Meta's 11B multimodal Llama 3.2 — the first vision-capable Llama model, supporting image understanding with 128K context.

Meta's ultra-compact 1B Llama 3.2 base model — designed for on-device and edge deployment with a 128K context window.

Meta's 1B Llama 3.2 instruction model — the smallest capable Llama, ideal for on-device chat with a 128K context window.

Meta's 3B Llama 3.2 base model — compact foundation for fine-tuning and on-device deployment with 128K context support.

Meta's compact 3B Llama 3.2 instruct model — capable chat and instruction-following in a lightweight, edge-friendly package.