Grok Imagine Image 2.0 API and Pricing, Explained Through a Real Pipeline

Grok Imagine Image 2.0 is xAI's image generation and editing model, the still-image sibling of the Grok Imagine video models. Give it a prompt and it makes a picture. Give it a prompt plus up to three reference images and it edits or composes from them. Output is 1K or 2K in thirteen aspect ratios. We put it on the Image Generation node and gave it a model page on October 14, 2026, after probing it through the same path the node uses.

Most of the questions people arrive with are about price, and this model has a rate card with more rows than most: two resolutions, two quality tiers, and a per-image charge for references. This post covers that rate card, what our test renders billed, which settings move the charge, how the reference edit behaves, and how to call Grok Imagine Image 2.0 as a published API endpoint instead of writing an xAI integration.

The spec sheet

The aspect ratio list is the broad one. Alongside the usual 1:1, 3:4, 4:3, 2:3, 3:2, 9:16 and 16:9, it adds 1:2 and 2:1, and the taller 9:19.5 and 9:20 shapes with their 19.5:9 and 20:9 mirrors, so a full-height phone frame and a wide header come out of the same node without a crop. Leave the ratio on auto and the model picks one from the prompt.

The last row is the one to plan around. The model returns one image per request, so a set of candidates is several runs on the node rather than one batch, and each run is billed on its own.

What Grok Imagine Image 2.0 costs per image

These are the listed prices as of October 2026. Two settings set the price, resolution and quality, and medium is the default quality:

Each input image adds $0.01, so an edit with all three references at 2K on medium is the ceiling at $0.11, and a plain 1K prompt on low is the floor at $0.04.

Two things in that table deserve a close read. Quality moves the price as much as resolution does: low to medium at 1K adds two cents, and so does 1K to 2K on low, so a 2K low render and a 1K medium render are the same $0.06. And the default is the dearer tier. If you never touch the quality setting, every 1K render lists at $0.06 and every 2K render at $0.08. Picking low is where the savings are, and it is a one-field change.

For scale: a hundred 1K drafts on low list at $4.00. The same hundred at 2K on medium list at $8.00.

What we measured on October 14

We probed the model on October 14, 2026, through the same path the Image Generation node uses, and the billed cost matched the rate card each time:

The text prompt asked for a ceramic cup of coffee on a sunlit wooden cafe table with steam rising and shallow depth of field. The result was a speckled stoneware cup on worn wood, steam curling up, a chair and a window plant softened behind it. 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 navy weave, the collar, the button and the fold, set it on veined white marble under soft light, and came in close on the fabric.

The charges break down against the rate card cleanly. The edit's $0.07 is $0.06 for the output, the medium 1K rate, plus $0.01 for the one input image. The plain prompt billed $0.04, the low 1K rate. A further probe the same day sent the quality setting explicitly on the same route: low billed $0.04 and the default billed $0.06, which is how we confirmed the provider honours the setting. With resolution left unset, the model returned a 1248 by 832 frame and billed the 1K price.

Speed was the standout. Both renders came back in well under the twenty seconds we have come to expect from the current crop of image models, and the edit was the faster of the two. Two renders on one day is not a benchmark, but it is quick enough to iterate on a prompt in one sitting.

Low to draft, medium to finish

With a two-tier rate card, the pattern is simple:

1. Draft on low at 1K. Explore framing and prompt wording at $0.04 a frame. Twenty-five variations list at $1.00.
2. Finish on medium when a frame earns it. Rerun the chosen prompt at medium, and at 2K if the frame is going anywhere larger than a phone screen. That is $0.08 once rather than twenty-five times.
3. Attach references on the finishing run. Each one is a cent, so leave them off while exploring composition and bring them in when placing the real product.

One detail on the node: the credit estimate shown before a run is quoted at the medium rate, $0.06 at 1K and $0.08 at 2K, plus a cent per reference. Set quality to low and the quote stays at the medium number, but the charge settles on what the provider billed, so a low render at 1K settles at $0.04 against its $0.06 quote. The number you see is a ceiling, not the bill.

Reference images and the edit path

With a prompt alone Grok Imagine Image 2.0 generates. With a prompt plus up to three references it edits or composes, on the same node. Wire the images into the Image Generation node's reference input, or drag them onto it, and write the change as the prompt: the same mug on a picnic blanket in a summer park, or the folded shirt on a marble counter.

Three references is a smaller ceiling than some of the catalog, where Nano Banana 2.1 takes up to 14 and FLUX 3 Image Klein up to 10. For a product, a setting, and a style it is enough. For a brief that has to carry a dozen assets into one frame, pick one of those instead.

Where it sits in the price band

Grok Imagine Image 2.0 is not the cheapest image model on the node, and it is worth being plain about that. Listed and measured prices as of October 2026 for the models around it:

What it offers over the cheaper rows is the turnaround we measured, 7 to 11 seconds against the roughly 20 seconds the Qwen tiers took in their own probes, and the thirteen-ratio list with its 9:20 and 20:9 extremes. If your frames are headed for a full-screen phone placement or a wide header and you want them back fast, it earns the extra cents. For a storyboard of fifty frames at 1:1, Seedream 5.0 Flash at under two cents a frame is the better default.

From a still to a clip: the Grok Imagine family

Grok Imagine Image 2.0 makes stills. Grok Imagine 1.5 and its distilled sibling Grok Imagine 1.5 Lite make video clips, and the common pipeline across the family is a two-node graph: make the key frame with the image model, then wire it into a video node's First frame port to animate it. The post on Grok Imagine 1.5 Lite pricing covers the video side of that pairing, setting a first-frame image for AI video generation covers the handoff, and the image to video generator page has the pairing set up.

What you actually pay on Treza

Grok Imagine Image 2.0 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. Treza quotes the price before a run and settles the charge on what the provider billed, which on this model means a low-quality render costs less than its quote.

Calling Grok Imagine Image 2.0 as an API

Direct access means xAI credentials, the request shape of the route this model runs on, error handling, and retry logic before you have built anything.

Through a Treza pipeline, Grok Imagine Image 2.0 is a node on a canvas. Place the Image Generation node, pick the model, set resolution, quality 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 to a different image model 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 frame 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 Grok Imagine Image 2.0 as the first node of several: compose the key frame, pass it to a video node as the first frame, then put narration, captions, or publishing behind the video. A fuller AI video pipeline puts a script step ahead of the image and 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, the AI thumbnail generator page covers the other end of the graph, and how to build an AI image generation pipeline walks through image graphs in more depth.

Frequently Asked Questions

How much does Grok Imagine Image 2.0 cost per image?

The listed prices as of October 2026 are $0.04 at 1K and $0.06 at 2K on low quality, and $0.06 at 1K and $0.08 at 2K on medium, the default. Each reference image adds $0.01. Our 1K text-to-image probe billed $0.04 and our one-reference edit billed $0.07, both matching the listed rates.

Does the quality setting change the price?

Yes. Low is two cents cheaper than medium at either resolution, and the provider honours the setting: in our probe, low billed $0.04 at 1K against $0.06 for the default. On the Treza node the estimate is quoted at the medium rate and the charge settles on what the provider billed, so a low render costs less than its quote.

Do reference images cost extra with Grok Imagine Image 2.0?

Yes, $0.01 per input image, up to three per request. Our edit with one reference at 1K billed $0.07, which is the $0.06 medium rate for the output plus a cent for the reference. Wire references into the Image Generation node's reference input and describe how they should be used.

What resolutions and aspect ratios does it support?

1K or 2K, in thirteen aspect ratios plus auto: 1:1, 3:4, 4:3, 2:3, 3:2, 9:16, 16:9, 1:2, 2:1, 9:19.5, 19.5:9, 9:20 and 20:9. With resolution left unset, our probe returned a 1248 by 832 frame at the 1K price.

How is Grok Imagine Image 2.0 different from Grok Imagine video?

Grok Imagine Image 2.0 makes stills. Grok Imagine and Grok Imagine 1.5 make video clips. A common pipeline makes a still with this model and wires it into a video node's First frame port to animate it.

Is Grok Imagine Image 2.0 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.