A few months after launching Qwen3-VL, Alibaba has released a detailed technical report on the open multimodal model. The data shows the system excels at image-based math tasks and can analyze hours of video footage.