Official image-text-to-text model by Qwen.

Official zero-shot-image-classification model by microsoft.