Alibaba Cloud just dropped Qwen-Image-2.0-Pro, the full-featured version of its image generation series, on its cloud platform Bailian (Chinese for "hundred refinements"). The model does something we don't see often: it bakes text-to-image generation and image editing into a single model. You can swap objects, rewrite text, or change the style of an image using plain language — no need to jump between tools.
Compared to the March 3rd version, the Pro update brings noticeable improvements in material detail, lighting, and texture rendering. Text rendering, previously limited to English, now supports multiple languages, tackling the garbled-text problem that plagues many AI image models. Semantic understanding is stronger too. Output resolution goes up to 2K. API pricing is $0.075 per image.
It's a similar all-in-one approach to OpenAI's ChatGPT Images 2.0, which also supports 2K output.