Qwen2.5-VL-3B-Instruct

Qwen2.5-VL-3B-Instruct is a 3-billion-parameter instruction-tuned vision-language model from Alibaba Cloud's Qwen team, capable of understanding images, documents, charts, and video. It is the smallest variant in the Qwen2.5-VL family, designed for efficient deployment across multimodal tasks including document analysis, visual question answering, structured output generation, and agentic UI interaction. Source: supabase.