MetaLlama 3.2
Llama 3.2 90B Vision Instruct
The 90B release is the larger Llama 3.2 image-and-text instruction model.
What the release provides
Its vision adapter uses cross-attention into the language model. It has 128K context and English image-task support under the Llama 3.2 Community License.Llama 3.2 vision model card
How to use it well
Practical guidance. Compare it with the 11B model on the exact diagrams and layouts your application receives. Review the smallest text and ambiguous relationships between objects. The larger checkpoint adds serving requirements that should be justified by task quality.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | meta-llama/Llama-3.2-90B-Vision-Instruct |