Qwen-Image is the Qwen team's family of open image models that both create pictures from a text prompt and edit existing ones.
In plain words
Qwen-Image turns a written prompt into an image and can also change an existing picture. It is known for clear, readable text inside images, Chinese in particular. The first weights came out in August 2025. Qwen-Image-2.1, released on 20 September 2026, does creation and editing in one model. It can make transparent backgrounds, use up to 10 reference pictures, and change only an area you circle or paint over. A Turbo version from 9 October 2026 needs only 8 denoising steps (rounds of cleaning noise out of the picture).
A quick example
Imagine a small bakery that needs a poster showing its name and opening hours in clear lettering. The owner describes the poster in a sentence and gets a draft. Then they circle the cake in the picture and ask for a different one, and only that part changes.
Where you'll meet it
- Posters, signs and other images that need readable text.
- Editing photos and product pictures.
Related
Sources
- DocsQwen/Qwen-Image, Qwen (Hugging Face) · read 9 Oct 2026
- DocsQwenLM/Qwen-Image-2.1, Qwen (GitHub) · read 9 Oct 2026
- DocsQwen/Qwen-Image-2.1-Turbo, Qwen (Hugging Face) · read 9 Oct 2026
- DocsQwen/Qwen-Image-2.1, Qwen (Hugging Face) · read 9 Oct 2026
Checked by WhatIsAI against the sources above · 9 Oct 2026. A full five-minute explainer with a diagram may follow. Spotted an error? Report it.