Alibaba released Qwen-Image-3.0 on July 21, saying the foundation model for image generation supports inputs of up to 4,500 tokens and can generate knowledge diagrams and complex user interface layouts in a single pass.

The company described the model as the third-generation foundation model in its Qianwen image generation and editing series.

Alibaba also said Qwen-Image-3.0 supports native rendering in 12 languages and more than 20 fonts.

🔗 Source: Yicai