Ali Qwen releases Qwen-Image-3.0, super-long commands allow complex graphic and text to be generated with a single click.

On July 21st, Alibaba released the latest image generation base model Qwen-Image-3.0. The new model supports a maximum of 4.5k token ultra-long input, capable of generating knowledge graphs that incorporate formula symbols, geometric shapes, logical deduction steps, and other elements all in one go. It also supports the native rendering of 12 languages and more than 20 fonts, resulting in clear text and significantly reducing the production costs of "readable and usable" commercial materials such as multilingual product posters, film/cartoon storyboards, etc.