Llama 3.2 vision vs. Molmo: Open multimodal models | Interconnects AI | Podwise