Qwen2-VL-7B-Instruct

Qwen2-VL-7B-Instruct is a 7-billion-parameter instruction-tuned vision-language model from Alibaba's Qwen team, capable of understanding images, long-form video, and multilingual text within visual content. It supports tasks such as document understanding, visual question answering, and agentic visual reasoning. Source: supabase.