Qwen 3.0 Image Generation and Editing
Qwen Image 3.0 Pro creates high-resolution images from text and rewrites one to three reference images from plain-language instructions, in one family.
- Alibaba
- 2 variants
- Price
- See price
- Speed
- ~15s
- Best for
- Image edit · Image to Image
- Price
- See price
- Speed
- ~1 min
- Best for
- Text to Image
More from Alibaba.
A family groups the versions of one model line. Open one to compare its variants side by side.
Working with Qwen 3.0.
Qwen Image 3.0 Pro is Alibaba's latest image pair on each::labs: one variant renders from a text prompt with optional negative prompts and explicit sizing, the other takes one to three reference images and applies natural-language edit instructions. Both can return several images per request.
When to choose Qwen 3.0
Choose Qwen 3.0 for the newest Qwen image quality and multi-reference editing. Qwen 2.0 is the better pick when in-image typography is the job, and the Qwen Image Edit family when you need angle control or LoRA-assisted retouching rather than generation.
Choosing between the variants
Text-to-Image starts from nothing and gives you negative prompts and size control. Image Edit is the one that accepts up to three references, so use it for product shots, composites and any change that has to preserve an existing subject.
Questions about Qwen 3.0
The short answers. For anything specific to your setup, talk to our team.
Talk to an engineerPlan for your expected volume.The Image Edit variant accepts one to three reference images in a single request and composes the instruction across all of them.
Yes. Both variants support multi-image output, which is the cheapest way to see several interpretations of a prompt before committing to one.
The Text-to-Image variant does. Use it to name what must not appear — extra limbs, watermarks, background clutter — instead of over-describing the subject.