Rich content
Use prompts of up to 4.5K tokens to generate complex newspapers, storyboards, menus, exam papers and nested image-within-image layouts in one pass.
Qwen
QwenQwen Image 3.0 Pro is Alibaba's limited-preview generation and editing model for complex layouts, small text, detailed photography and multilingual visual content. It accepts up to three reference images and currently requires Model Gallery approval.
New · limited preview
Qwen's third-generation image model combines long-instruction layout control, fine text and photographic detail with multilingual knowledge and image editing.
Use prompts of up to 4.5K tokens to generate complex newspapers, storyboards, menus, exam papers and nested image-within-image layouts in one pass.
Render text as small as 10px while preserving fine visual texture such as pores, micro-expressions and individual strands of hair.
Create in 12 languages with broad font coverage, or simulate familiar web, game and livestream interfaces using the model's world knowledge.
Alibaba currently lists qwen-image-3.0-pro as a limited preview requiring Model Gallery approval. Pricing is not yet listed in the international pricing table.
Release details verified July 26, 2026.
Output

Estimate shown; confirm final terms in Alibaba Cloud.