Qwen-Image-2.1-Turbo
Qwen released Qwen-Image-2.1-Turbo, an accelerated checkpoint of Qwen-Image-2.1 that does text-to-image generation and editing in 8 denoising steps.
Turbo uses the same 7B visual generation architecture as Qwen-Image-2.1 but runs in 8 denoising steps, with the sampling schedule stored in the checkpoint. It uses CFG=1 by default and prefix KV caching that reuses text and reference-image context across steps. Hosted Pro and Turbo APIs went live on Alibaba Cloud Model Studio the same day.
- Date
- Friday 9 October 2026
- Lab
- Alibaba (Qwen)
- Kind
- open-weights
- Access
- open weights (restricted license)
Figures
| Measure | Value | Measured by |
|---|---|---|
| Denoising steps | 8 Turbo checkpoint, for both generation and editing | company |
| Visual generator parameters | 7B same architecture as Qwen-Image-2.1 | company |
Weights are under the Qwen research license, which Kingy AI says allows noncommercial research and evaluation only, with commercial use requiring a separate license. Kingy AI also says Alibaba Cloud publishes a Turbo API price.
Sources
- huggingface.co/Qwen/Qwen-Image-2.1-Turbo
- github.com/QwenLM/Qwen-Image-2.1
- kingy.ai/blog/qwen-image-2-1-turbo-eight-steps-license/
This record was checked against its sources on 9 October 2026. How we check