Updated

Qwen-Image-2.1

Qwen released Qwen-Image-2.1, an open-source image model that combines text-to-image generation and image editing in a single pipeline with native transparent image support and multi-image editing.

By No Code MBA · Facts checked

Provider
Qwen
Announcement date
Status at last check
Generally available
Model family
Qwen-Image
Capabilities
Image

What changed

Qwen-Image-2.1 combines text-to-image generation and image editing in one model with a 7B-parameter, 32-layer DiT generation component. It folds in transparent RGBA generation previously handled by Qwen-Image-Layered, can edit transparent layers, extract subjects from photos, take up to 10 reference images, and apply local edits marked by circles, annotations or masks. Outputs reach 2048x2048 at 1:1, and Qwen says KV cache reuse improves speed and memory use.

Access and availability

Weights are on Hugging Face at Qwen/Qwen-Image-2.1 and run through the Diffusers library.

Official access documentation

What you can build

Try generating a product image with an alpha channel in a no-code design tool, then edit it locally using reference images in one model workflow.

Limitations

The weights use the Qwen Research License rather than the Apache 2.0 license of earlier Qwen-Image models. Review the license terms before commercial use. Qwen published benchmark results only as charts and lists no known limitations.

Sources and release history