Qwen Image2.1✳
A new frame of possibility.
From the smallest details to the biggest ideas. A first look at Qwen’s next image generation and editing model.
An early look, straight from the source.
See what’s taking shape.
Real previews from the people trying it first.
Every frame has a creator. Every creator has a link.
46 examples / Early access collection
There’s more to the picture.
A collection of public early-access examples, attributed to their original creators. Results are creator-reported previews, not an independent benchmark or a complete record of every generation.
UPDATED SEP 19, 2026Imagine it.
Then make it yours.
Generation and editing, brought together.
Here’s what the public previews and technical
documentation tell us so far.
Start with an idea.
Turn a text prompt into a new image. Explore a range of styles, compositions and visual details.
Pipeline documentationKeep what matters.
Use one or several reference images to guide an edit. Change the scene, revisit a character, or try a new direction.
Image conditioningLet the image speak.
Early examples explore dense infographics, menus and posters. The Comfy preview lists native 2K and alpha-channel support.
Read the model previewCreate with
Qwen Image 2.1.
Find your inspiration here. Then take it into a space of your own — shape a prompt, explore a direction, and prepare your next image.
Good work deserves credit.
26 original showcase posts. 11 creators.
Plus the primary sources behind the model details.
The early-access announcement
A preview program for 50 creators and developers, ahead of the public release.
Inside the new image pipeline
Text-to-image and image-conditioned editing, including multiple reference images.
What’s coming to the ecosystem
The preview lists native 2K output, typography and alpha-channel support.



