Qwen Image 2.1 API and Pricing, Explained Through a Real Pipeline
Qwen Image 2.1 is Alibaba's image generation and editing series in two tiers that share one spec sheet: Pro for quality, Turbo for price. Both list a flat price per image at 1K and 2K, take up to 4 reference images for free, and came back in about 20 seconds when we probed them. Here is the rate card, what we measured on the day they landed, where each tier fits, and how to reach them through a published API endpoint.
Qwen Image 2.1 is the image generation and editing series from Qwen, Alibaba's model family. It reached OpenRouter on October 17, 2026, in two tiers, qwen/qwen-image-2.1-pro and qwen/qwen-image-2.1-turbo, and we added both to the Image Generation node, to chat, and to the model pages the same day. Each tier generates from a prompt alone, and edits or composes from up to 4 reference images.
The two tiers take identical inputs and settings, so the choice between them comes down to price and speed, and the price is the simpler of the two. This post covers the rate card, what we measured when we onboarded the models, how they compare with Qwen Image 3, and how to call either tier as a published API endpoint.
The spec sheet
| Capability | Qwen Image 2.1 Pro | Qwen Image 2.1 Turbo |
|---|---|---|
| Resolution tiers | 1K, 2K | 1K, 2K |
| Aspect ratios | 13, from 1:4 through 4:1 | 13, from 1:4 through 4:1 |
| Input | Text and image | Text and image |
| Reference images | Up to 4 per request | Up to 4 per request |
| Seed | Yes | Yes |
Every row is the same on both sides, so a pipeline built on one tier switches to the other by changing the model on the node and nothing else. The aspect ratio list runs from 1:4 and 4:1 strips through 9:16, 4:5, 1:1, and 16:9, so one node covers the vertical, feed, and landscape sizes a channel needs. The full list is on the Qwen Image 2.1 model page.
What Qwen Image 2.1 costs per image
These are the listed prices as of October 2026. Both tiers charge a flat price per image, the same at 1K and at 2K, and neither charges for input images:
| Tier | Per image at 1K | Per image at 2K | Reference images |
|---|---|---|---|
| Qwen Image 2.1 Pro | $0.046 | $0.046 | Free, up to 4 |
| Qwen Image 2.1 Turbo | $0.0184 | $0.0184 | Free, up to 4 |
Three things in that table are worth reading closely.
Resolution is free. Most image models on the node price 2K above 1K. Qwen Image 2.1 does not. If a frame is going anywhere larger than a phone screen, pick 2K and pay the same.
Turbo is two and a half times cheaper than Pro. A hundred images list at $4.60 on Pro and $1.84 on Turbo. Twenty 2K renders list at $0.92 on Pro and $0.37 on Turbo.
References cost nothing extra on either tier. An edit with four references attached is billed the same as a plain prompt, and the credit estimate on the node does not move when you attach them.
On the node, leaving Resolution on Model default renders a 1024 by 1024 square at the same flat price. Pick 2K explicitly when you want it.
What we measured on launch day
We probed each tier three times on October 17, 2026, the day they landed, through the same path the Image Generation node uses. The billed cost matched the listed price in every case:
| Tier | Probe | Setting | Billed | Wall time |
|---|---|---|---|---|
| Pro | Text to image | 1K, 1:1 | $0.046 | about 20 seconds |
| Pro | Text to image | Resolution unset, 1:1 | $0.046 | about 23 seconds |
| Pro | Edit with one reference | 1K, 1:1 | $0.046 | about 22 seconds |
| Turbo | Text to image | 1K, 1:1 | $0.0184 | about 18 seconds |
| Turbo | Text to image | Resolution unset, 1:1 | $0.0184 | about 20 seconds |
| Turbo | Edit with one reference | 1K, 1:1 | $0.0184 | about 20 seconds |
The unset runs confirmed the default: a 1024 by 1024 square on both tiers. The reference runs used a product still of a folded navy linen shirt and asked for the same shirt on a white marble counter in a bright studio. Both tiers kept the shirt, its color, fold, collar and buttons, and placed it on marble under soft daylight. Pro framed the whole garment; Turbo came in closer on the collar and the weave. Neither cost more than a plain prompt.
Speed was the surprise. Turbo is the tier named for it, but across six renders the two tiers finished within a few seconds of each other, both near the 20 second mark. Six renders from one day is a small sample, not a benchmark, but it suggests the gap between the tiers is mostly a price gap.
Pro or Turbo
With both tiers on the node, the pattern that falls out of the rate card is simple:
- Draft on Turbo. Explore composition, framing, and prompt wording at $0.0184 a frame. Fifty variations list at $0.92.
- Batch on Turbo. Storyboards, carousel frames, thumbnail candidates, and product sets are volume jobs, and a tier under two cents a frame fits them.
- Finish on Pro when a frame earns it. Run the chosen prompt and references through Pro on a second node. If it looks better for your subject, pay $0.046 once rather than fifty times.
We are not going to tell you which tier looks better, because that depends on your subject. Put the same prompt and references through both on two nodes and compare.
How Qwen Image 2.1 compares with Qwen Image 3
Qwen Image 3 is the other Qwen series on the node. Both generate and edit at 1K or 2K, with the same thirteen aspect ratios and up to 4 references. Qwen Image 3 is built for precise small text, down to 10 pixels, so menus, labels, and small print hold up in the frame. It is also slow enough that the node gives it extra time to finish, which is why Qwen Image 2.1 is the series to reach for in chat when you want a still back quickly.
On price, Qwen Image 3 lists at $0.03 per image plus $0.003 per reference, and Qwen Image 3 Pro at $0.04 at 1K and $0.075 at 2K, also plus $0.003 per reference. Qwen Image 2.1 Turbo undercuts both, and Qwen Image 2.1 Pro at 2K costs less than Qwen Image 3 Pro at 2K. If your frames carry legible text, test Qwen Image 3 first. If they do not, Turbo is the cheaper default from the same lab.
Two models from other labs sit in the same price band. Seedream 5.0 Flash lists at $0.018 flat per image with up to 14 references, and FLUX 3 Image Klein at $0.009 at 1K and $0.014 at 2K with up to 10. Qwen Image 2.1 is not the cheapest image model on the node. What it offers is a Pro and Turbo pair on one spec sheet, and the 1:4 and 4:1 strip formats most of the catalog does not have.
Reference images and consistent sets
With a prompt alone Qwen Image 2.1 generates. With a prompt plus references it edits or composes, on the same node. Attach a product, a setting, and a style reference, then ask for one composition, and the result carries all three forward.
Because references are free and the price is flat, generate several candidates per shot on Turbo, keep the one that matches best, and move on. Fix the seed and vary the prompt to keep a series closer together. The cleaner the reference still, the less the result drifts.
The endpoint detail that trips up integrations
If you call Qwen Image 2.1 through OpenRouter yourself, note that both tiers run on the images endpoint only. When we probed them, the chat completions route refused the request on Pro and on Turbo alike, the same as the FLUX 3 family and Qwen Image 3. An integration written for chat-completions image output needs a second code path for these models.
Inside a Treza pipeline you never see this. The Image Generation node knows which models require the images endpoint and routes the request there. You pick Qwen Image 2.1 Pro or Turbo in the model menu and the node handles the rest. In chat, ask for either tier by name and the image renders in the conversation.
What you actually pay on Treza
Qwen Image 2.1 through a Treza pipeline is metered from a prepaid credit balance. One credit equals $0.01, packs start at $5, credits never expire, and there is no subscription. A failed generation charges nothing, and that includes a prompt the provider refuses.
The credit estimate on the node uses the listed flat price for the tier you pick, so the number you see before a run is the number you pay for a successful render. Turbo rounds to two credits an image. Pro rounds to five.
Calling Qwen Image 2.1 as an API
Direct access means provider credentials, the images endpoint request shape, error handling, and retry logic before you have built anything.
Through a Treza pipeline, Qwen Image 2.1 is a node on a canvas. Place the Image Generation node, pick Pro or Turbo, set resolution and aspect ratio, wire references into its reference input, and publish the graph as a single versioned endpoint. Your app calls the typed /invoke route with JSON in and JSON out, or points an existing OpenAI SDK at the OpenAI-compatible route. Swap the node from Turbo to Pro later and the endpoint your product calls stays the same.
The same pipeline is reachable from an agent. Treza's hosted MCP server exposes pipelines as tools that Claude and other MCP clients call directly, so an agent can generate the image and hand back the URL as one step of a larger task.
Putting it in a pipeline instead of a single call
A single image is rarely the deliverable. The more useful pattern is Qwen Image 2.1 as the first node of several.
Compose a key frame from your references, then pass it to a video node as the first frame. The post on setting a first-frame image for AI video generation covers the handoff, the image to video generator page has that pairing set up, and the AI thumbnail generator page covers the other end of the graph, where a thumbnail is built from the finished video.
A fuller AI video pipeline puts a script step ahead of the image and narration, captions, or publishing behind the video, then reruns the whole graph on a schedule or from an API call. The AI image generator page has the image side set up, the AI video generator page puts a video model behind a single generate action, and how to build an AI image generation pipeline walks through image graphs in more depth.
Frequently Asked Questions
How much does Qwen Image 2.1 cost per image?
The listed prices as of October 2026 are $0.046 per image on Qwen Image 2.1 Pro and $0.0184 per image on Qwen Image 2.1 Turbo. The price is flat: 1K and 2K cost the same, and aspect ratio and reference images do not change it.
What is the difference between Qwen Image 2.1 Pro and Turbo?
They take the same inputs: a prompt, up to 4 reference images, 1K or 2K output, thirteen aspect ratios, and a seed. Pro is the higher quality tier at $0.046 per image; Turbo is the lower priced tier at $0.0184. In our six probes on launch day both finished in about 18 to 23 seconds, so the practical difference is price and output quality, which only a side-by-side run of your own prompt can settle.
Do reference images cost extra with Qwen Image 2.1?
No. Both tiers accept up to 4 reference images per request and list no charge for them. When we probed them, an edit with one reference billed the same flat price as a plain text prompt on each tier.
How is Qwen Image 2.1 different from Qwen Image 3?
Both generate and edit at 1K or 2K, with the same thirteen aspect ratios and up to 4 references. Qwen Image 3 is built for precise small text, down to 10 pixels, and renders more slowly. Qwen Image 2.1 is the faster series and the one to use in chat when you want a still back quickly. Qwen Image 3 also charges $0.003 per reference, where Qwen Image 2.1 charges nothing for them.
Does Qwen Image 2.1 work through the chat completions API?
Not on OpenRouter as of October 2026. Both tiers run on the images endpoint, and the chat completions route refused them when we tested. On Treza the Image Generation node routes the request to the right endpoint automatically, and both tiers are available by name in chat.
Is Qwen Image 2.1 free?
Building a pipeline on the canvas costs nothing, but generation spends real provider money and is charged from a prepaid credit balance. There is no subscription, packs start at $5, credits never expire, and a failed generation charges nothing.
