Decision summary
- Up to 10 reference images or native transparency -> 2.1 Pro
- 120 RPM or open weights -> 2.1 Turbo
- Singapore billing -> use the Singapore per-image price
Image model comparison
Qwen-Image 2.1 Pro vs Qwen-Image 2.1 Turbo
Qwen-Image-2.1 Pro, released October 2, 2026, is the hosted image API ID qwen-image-2.1-pro at about $0.035 per image and 20 requests per minute, with up to 10 reference images.
Featured · Updated today · Last verified: October 2026
Choose Qwen-Image 2.1 Pro when
Hosted image generation and editing when reference images and transparency matter.
Choose Qwen-Image 2.1 Turbo when
Higher-volume hosted calls, or running the open weights yourself.
- Up to 10 reference images or native transparency: 2.1 Pro
- 120 RPM or open weights: 2.1 Turbo
- Singapore billing: use the Singapore per-image price
Decision axes: Best for · Access · Context · List price
How they compare
Criterion-by-criterion notes from the catalog—not a ranking. Validate on your own gold set.
| Criterion | Qwen-Image 2.1 Pro | Qwen-Image 2.1 Turbo |
|---|---|---|
| Best for | Hosted image generation and editing when reference images and transparency matter. | Higher-volume hosted calls, or running the open weights yourself. |
| Access | API ID qwen-image-2.1-pro. Released October 2, 2026. | API ID qwen-image-2.1-turbo. Open weights. Released October 9, 2026. |
| Context | 7B visual component. Up to 10 reference images. Native transparency. | 8-step generation. |
| List price | $0.035333 per image in Beijing, the US, Germany, Japan, and Hong Kong. Singapore is $0.04. 20 requests per minute. | $0.014133 per image, or $0.016 in Singapore. 120 requests per minute. |
| Controls | Hosted API. Not the open-weight Turbo checkpoint. | Open weights if you self-host. The API price is the hosted path. |
| Operational risk | The low rate limit is the constraint before the per-image price is. | Do not assume Turbo matches Pro on reference-image count or transparency. Those are documented on Pro. |
Key insights
Concrete technical or product signals.
- Turbo is roughly 40% of Pro's listed per-image price and six times the request rate.
- Qwen-Image-3.0 stays a separate generation in the catalog.
Use cases
Where this shines in production.
- Picking a Model Studio image ID
- Trading quality headroom for throughput
Limitations & trade-offs
What to watch for.
- No side-by-side quality score is stored here.
- Regional prices differ. Quote the region you will call.
Who should not choose this?
Signals that this pair is the wrong default.
- Do not use one region's price in another region.
- Do not retire Qwen-Image-3.0 because 2.1 shipped.
- Do not self-host Pro's API ID. The open weights are Turbo.
Final recommendation
What to do after reading the table.
Default high-volume jobs to Turbo and send reference-heavy jobs to Pro.
FAQ
Is Qwen-Image 2.1 Pro better than Qwen-Image 2.1 Turbo?
Default high-volume jobs to Turbo and send reference-heavy jobs to Pro.
Which is cheaper: Qwen-Image 2.1 Pro or Qwen-Image 2.1 Turbo?
Listed unit prices: Qwen-Image 2.1 Pro $0.035333 per image in Beijing, the US, Germany, Japan, and Hong Kong. Singapore is $0.04. 20 requests per minute; Qwen-Image 2.1 Turbo $0.014133 per image, or $0.016 in Singapore. 120 requests per minute. A lower listed unit price is not the same as a lower total workload cost—tokens, session minutes, caching, introductory rates, and add-on modes can reverse the ranking. Confirm live rates on each vendor’s official page.
Can I use both Qwen-Image 2.1 Pro and Qwen-Image 2.1 Turbo?
Yes. Many teams keep both and route by constraint. Use Qwen-Image 2.1 Pro for Hosted image generation and editing when reference images and transparency matter, and Qwen-Image 2.1 Turbo for Higher-volume hosted calls, or running the open weights yourself. Confirm vendor terms before mixing them in one production workflow.