Grok Imagine Image 2 does not yet have a separately documented API model ID. xAI's August 7, 2026 consumer launch says Image 2.0 API access is coming soon. At the same time, xAI already operates an Imagine API with two documented image models: grok-imagine-image and grok-imagine-image-quality.
Do not silently map the new consumer Quality Mode to the existing quality API alias. The names look related, but xAI has not published that identity claim in the sources reviewed on August 9.
Use Oakgen's current Grok Imagine model for the route available today, or read the Image 2.0 launch guide for the consumer features.
Current API status
| Item | Official status | What it means |
|---|---|---|
| Image 2.0 consumer mode | Live | Quality Mode on grok.com and mobile |
| Exact Image 2.0 API ID | Not documented | Launch page says coming soon |
| grok-imagine-image | Live API model | $0.02 output at 1K or 2K |
| grok-imagine-image-quality | Live API model | $0.05 at 1K; $0.07 at 2K |
| Editing references | Up to 3 in current API docs | Consumer Image 2.0 says up to 5 |
| Rate limit | 5 requests/second | Listed for both image models |
Current xAI image prices
For grok-imagine-image, xAI lists $0.002 per image input and $0.02 per generated image at either 1K or 2K.
For grok-imagine-image-quality, it lists $0.01 per image input, $0.05 per 1K output, and $0.07 per 2K output. Image generation uses flat output pricing rather than prompt-token pricing.
Those numbers can change. Read the live xAI pricing page before calculating customer charges or credit packs.
Available operations in the existing Imagine API
xAI's developer overview documents text-to-image generation, natural-language image editing, up to three reference images per edit, 1K and 2K output, and up to ten generated outputs per request. The image endpoints follow OpenAI-compatible patterns for generation and edits.
The platform also supports Files API references, which can keep repeated source images private and avoid uploading the same input for every edit.
Example request for the documented quality model
curl -X POST https://api.x.ai/v1/images/generations \
-H "Content-Type: application/json" \
-H "Authorization: Bearer $XAI_API_KEY" \
-d '{
"model": "grok-imagine-image-quality",
"prompt": "A clean editorial poster for a trail shoe, exact headline: RUN LIGHT",
"n": 1
}'
This calls the existing documented quality model. It is not proof that the response uses the August 7 consumer Image 2.0 release.
Production checklist
- Fetch the model list for the authenticated account instead of hard-coding an undocumented ID.
- Store generated files immediately; provider URLs may not be permanent.
- Add retry behavior for 429 and server errors, but cap attempts and log the final status.
- Track input-image and output-image charges separately.
- Record the exact model alias used with every output.
- Test reference count and resolution against the live endpoint.
For prompts and edit patterns, see the Grok prompt library and editing guide.
Consumer pricing is a different question
Grok is free to start, while paid SuperGrok plans raise limits through a shared weekly allowance. That does not prove unlimited Image 2.0 use, and the public docs reviewed here do not give a simple per-image consumer price.
Avoid pages that turn an API rate into a claim about a Grok subscription. They are different billing systems.
Where Oakgen fits
Oakgen's current Grok Imagine route gives creators a UI alongside other image models and editing tools. It uses Oakgen's credit system rather than asking the user to manage an xAI API key. If the exact Image 2.0 API model becomes available, the model ID and capabilities should be verified before Oakgen or any other platform labels the integration.
Common mistakes
- Inventing
grok-imagine-image-2as an API ID. - Calling the existing quality alias Image 2.0 without a source.
- Copying the five-reference consumer limit into API validation.
- Forgetting the input-image fee on edits.
- Treating temporary output URLs as permanent storage.
Frequently asked questions
What is the standard model ID?
grok-imagine-image is the current standard ID in xAI's docs.
What is the quality model ID?
grok-imagine-image-quality is the current documented quality ID.
How much is a 2K quality image?
xAI lists $0.07 per 2K output, plus $0.01 for each image input used by an edit.
Can the API edit multiple images?
Yes. The current docs say up to three source images in one edit request.
When will Image 2.0 get API access?
xAI says it is coming soon but gives no date in the August 7 announcement.
