Meta releases Llama 3.2 with vision and lightweight models

Llama 3.2 includes 11B and 90B multimodal vision models plus lightweight 1B and 3B text models for edge devices.