Llama 3.2 11B Vision

Llama 3.2 11B Vision is Meta’s multimodal model for working with both images and text. It can answer questions about an image and generate language from visual context; calling it merely an 11B text model misses its defining capability. Multimodal means the model accepts more than one kind of input, here vision plus language. Its language component builds on the Llama family while an image-processing component connects visual information to it. TT-Metal lists this model with TP=2 on an n300 Wormhole system and batch 16. That is a two-device tensor-parallel configuration, smaller in chip count than the QuietBox deployments used for many 32B and 70B text models in the matrix.