Alibaba launches AI model which transforms screenshots into code
Alibaba has launched its large language model (LLM) AI model, Qwen3-VL, which is an open-sourced 235-billion-parameter vision-language model excelling in visual question-answering, GUI navigation, and video processing. It is available in Instruct and Thinking versions, with the former version outperforming Gemini 2.5 Pro on key vision benchmarks. It also transforms screenshots into code.