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.