Riverflow 2.6 Pro API and Pricing, Explained Through a Real Pipeline
Riverflow 2.6 Pro is the flagship of Sourceful's Riverflow 2.6 family: one model for text to image and image to image, 1K to 4K, up to 10 reference images, and a real transparent background on request. OpenRouter lists 13, 15 and 17 cents per image, and our tests billed a little more. Here is what a render actually cost and took, what came back, why the transparent cutout only works on one route, and how to reach the model through a published API endpoint.

Riverflow 2.6 Pro is the flagship of Sourceful's Riverflow 2.6 family, and it is built for the product end of image work: a label on three bottle shapes, a folded shirt moved onto a new counter, a cutout with a real alpha channel ready to drop on a layout. It reached OpenRouter on October 29, 2026, and we put it on the Image Generation node and on a model page the next day, after probing it through the same path the node uses.
The pricing question has a wrinkle: OpenRouter lists a clean rate card, and the model bills a little above it. This post covers the spec sheet, what three test renders billed against the list and how long they took, the one setting that only works on one route, and how to call the model as a published API endpoint.
The spec sheet
| Capability | Riverflow 2.6 Pro |
|---|---|
| Input | Text and image |
| Resolution | 1K, 2K or 4K |
| Aspect ratios | 10, from 9:16 to 21:9, plus auto |
| Reference images | Up to 10 |
| Background | Transparent or opaque, PNG output |
| Output formats | PNG, JPEG, WebP |
| Batch per node | 1 to 4 images, each charged separately |
The ten ratios are 1:1, 4:3, 3:4, 3:2, 2:3, 5:4, 4:5, 16:9, 9:16 and 21:9. Two of them, 5:4 and 4:5, are new against Riverflow 2.5 Pro, and 4:5 is the one a social feed post needs. The rest of the sheet matches 2.5 Pro: up to 4K, up to 10 references, and a transparent background on request.
One model covers both directions. A prompt alone generates. A prompt plus reference images edits or composes on the same node, with the change written as the prompt: this label on these three bottle shapes, the same sneaker on a pastel podium at 4:5. Wire the images into the node's reference input or drag them onto it, up to ten.
What Riverflow 2.6 Pro costs per image
As of October 2026 OpenRouter lists Riverflow 2.6 Pro per image by resolution:
| Resolution | Listed per image |
|---|---|
| 1K | $0.13 |
| 2K | $0.15 |
| 4K | $0.17 |
Those are the same three numbers as Riverflow 2.5 Pro. What the model actually bills is the reason this post exists. Our three probes on October 30, 2026, all at 1K and 1:1:
| Probe | Route | Billed | Wall time |
|---|---|---|---|
| Text to image | chat completions | $0.152 | 87 seconds |
| Edit with one reference | chat completions | $0.168 | 141 seconds |
| Transparent PNG cutout | images endpoint | $0.154 | 78 seconds |
Every render billed above the 13 cent list price. The provider reports each image as 4,175 output image tokens, and the charge lands around 15 cents whichever route it takes. The reference edit cost about a cent and a half more, so references carry a small premium rather than being free. Three renders is not a benchmark, but all three landed on the same side of the list, and Riverflow 2.5 Pro does the same, billing about 25 cents at 1K against the same 13 cent listing. Model a budget on the measured numbers.
The wall times matter too. A minute and a half for a prompt and over two minutes for an edit is slow for an image model, and it is why Riverflow 2.6 Pro is a canvas model on Treza rather than a chat one. It sits in the Image Generation node's model picker, where a long render does not hold up a conversation turn. In chat, Riverflow 2.5 Fast is the Riverflow to ask for, and it comes back in under half a minute.
What came back
The text prompt asked for a ceramic cup of coffee on a sunlit wooden cafe table, steam rising, shallow depth of field. The render is a cream stoneware cup with a blue-grey band, steam lit from the window, hard sunlight across the table, and the cafe behind it dissolved into bokeh. Everything the prompt named is there.
The edit took a product still of a folded navy linen shirt and asked for the same shirt on a white marble counter in a bright studio. The model kept the shirt, its collar, buttons and weave, laid it on veined marble, and built a bright room around it in soft focus. The product survives and the set changes.
The cutout probe asked for a ceramic coffee cup on a transparent background, with PNG output. It came back as a 1024 pixel square PNG with a real alpha channel: a plain white cup with soft shading, no background, and no checkerboard painted into the pixels. That last point is the one to check with any model that claims transparency.
Transparent output only works on one route
Riverflow 2.6 Pro can be called two ways through OpenRouter: over chat completions, which is how most image models run, and over the images endpoint. Chat completions drops the background and output format settings. We found this on Riverflow 2.5 Pro on October 14, 2026: a transparent PNG asked for over chat came back as an opaque WebP with a checkerboard drawn in, while the same request on the images endpoint returned an RGBA PNG with real alpha. Riverflow 2.6 Pro behaves the same way.
On the Treza node you do not have to know this. Pick Transparent in the Image Generation node's Background setting and the node routes that step to the images endpoint, which honours it. Leave the setting alone and the step runs over chat completions as usual. Riverflow 2.5 Fast is left out, since it outputs JPEG only, which has no alpha channel.
A real cutout is useful downstream. The Canvas node puts a picture on a solid colour as a card, and a transparent cutout sits straight on the colour with the shadow and rounded corners turned off. The Overlay node lays an element over a video, and its Transparent blend is for exactly this kind of alpha channel. A cutout on a Canvas card is an end card for a product video with no masking step.
Where it sits in the price band
Per image at 1K as of October 2026, listed where the provider publishes a rate and measured where we probed it:
| Model | Per image at 1K |
|---|---|
| Riverflow 2.5 Pro (measured) | $0.25 |
| Riverflow 2.6 Pro (measured) | $0.152 |
| FLUX 3 Image | $0.048 |
| Seedream 5.0 Pro | $0.045 |
| Nano Banana 2.1 (measured) | $0.036 |
| Recraft V4.1 | $0.035 |
| Riverflow 2.5 Fast (measured) | $0.0305 |
Riverflow 2.6 Pro sits near the top of the band, and the case for it is specific: reference-driven product finals and cutouts, where ten references, 4K and a real alpha channel are worth fifteen cents and a minute and a half. For a thumbnail or a first frame nobody will inspect at 4K, Seedream 5.0 or FLUX 3 Image Klein do the job for a few cents, and in chat. Against its own sibling, 2.6 Pro is the obvious pick: it costs less than 2.5 Pro, renders faster, and adds the two feed ratios.
What you actually pay on Treza
Riverflow 2.6 Pro 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. Building a pipeline on the canvas costs nothing either; only generation spends.
Before a run starts, the platform estimates the media cost of the graph and refuses to start one your balance cannot cover. For Riverflow 2.6 Pro the estimate follows the measured cost rather than the list price: 16 cents at 1K, 18 cents at 2K, 20 cents at 4K, plus two cents per reference image, and a node with the resolution left unset is quoted at the 2K rate. The charge settles on what the provider billed.
Calling Riverflow 2.6 Pro as an API
Through a Treza pipeline, Riverflow 2.6 Pro is a node on a canvas. Place the Image Generation node, pick the model, set the resolution, ratio and background, wire references in if you have them, 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 to a cheaper model for drafts 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 render the product still 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 pattern that suits Riverflow 2.6 Pro is a product line: a File node with the product shots, an Image Generation node on Riverflow 2.6 Pro with the shots wired in as references, a second image step for the transparent cutout, a Canvas node to put the cutout on a brand colour, and a product video behind it with the still as the first frame. A fuller AI video pipeline puts a script step ahead of the image and narration, captions and 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 Riverflow 2.6 Pro cost per image?
OpenRouter lists 13, 15 and 17 cents per image at 1K, 2K and 4K as of October 2026. Our tests on October 30, 2026 billed a little more: $0.152 for a 1K image from a prompt, $0.168 with one reference image, and $0.154 for a transparent PNG. Treza quotes the measured amounts before a run, and a failed generation costs nothing.
Can Riverflow 2.6 Pro return a transparent background?
Yes, with a real alpha channel, but only on the images endpoint; over chat completions the background setting is dropped. On the Treza node, pick Transparent in the Background setting and the step is routed to the images endpoint automatically. Our test cutout came back as a 1024 pixel square RGBA PNG in 78 seconds.
How is Riverflow 2.6 Pro different from Riverflow 2.5 Pro?
Both go up to 4K, take up to 10 references and can return a transparent background. 2.6 Pro adds the 5:4 and 4:5 ratios, rendered a 1K prompt in 87 seconds against about two minutes on 2.5 Pro, and billed about 15 cents at 1K against about 25 cents. The list price is the same on both.
Can Riverflow 2.6 Pro edit an image I already have?
Yes. Attach the image as a reference on the Image Generation node and write the change as the prompt. In our probe it kept a folded navy linen shirt intact and moved it from its original still onto a white marble counter in a bright studio, for $0.168.
Can I use Riverflow 2.6 Pro in Treza chat?
No. Its renders take a minute or two, so it runs on the canvas, in the Image Generation node, inside pipelines that can be published as API endpoints and MCP tools. In chat, ask for Riverflow 2.5 Fast, which comes back in under half a minute.


