On August 5, the Alibaba Qwen image generation model Qwen-Image-3.0 was officially launched on the Qwen AI platform and fully opened to users. The flagship Pro version and standard Standard version simultaneously opened their APIs, with the starting price for text-to-image generation at 0.18 yuan per image. Overseas users can access it through Qwen Cloud. In the latest text-to-image ranking on the international evaluation platform Arena.ai, this model ranked first domestically.

QQ20260805-113117.jpg

The Qwen-Image-3.0 focuses on ultra-long text understanding and detailed rendering. The model supports up to 4.5k token instructions, enabling the generation of complex layouts such as storyboards, exams, and nine-grid knowledge maps in one go. In terms of detail expression, it can precisely restore small text of 10px and entire pages of LaTeX academic formulas, with realistic skin and hair on human figures, and supports seamless restoration of ancient paintings. Meanwhile, the model covers over 100 art styles, native rendering in 12 languages, and includes world knowledge and online search capabilities.

QQ20260805-113125.jpg

Currently, the model has been integrated into commercial workflows such as e-commerce visual effects, film storyboards, advertising posters, and architectural rendering. In terms of pricing, the Standard version is 0.18 yuan per image for 1K/2K; the Pro version is 0.25 yuan and 0.5 yuan per image for 1K and 2K respectively, with image editing input as low as 0.02 yuan per image, further enhancing the efficiency and cost-effectiveness of commercial-grade AI image generation.

Qwen AI Platform:

https://platform.qianwenai.com/try-ai?scene=image&models=qwen-image-3.0-pro