Qwen mark

Qwen-Image

Qwen-Image is a 20B MMDiT text-to-image model that’s strongest at producing cohesive, good-looking images with reliable composition and generally solid physical logic. In portrait and selfie-style prompts it can deliver realistic faces and pleasing natural lighting, making it useful for “reference photo” generation (e.…

Speed

Price

Inputs

Text

Outputs

Image

Lab

Alibaba Cloud

Qwen-Image sample outputs

Captured directly from our eval suite. Click any tile to inspect the full render.

More from the Qwen family

Explore other options from the same family.

Launch your first agent today

Build without code. Share without limits.