Evaluating performance on extensive human evaluations and benchmarks, results suggest that Llama 3.2 vision models are competitive with leading closed models on image recognition + a range of visual understanding tasks.

→ View original post on X — @aiatmeta,