# Grok Imagine Image 2.0

SpaceXAI's image model for work that has to survive editing. Grok Imagine Image 2.0 generates from text, edits specific regions on request, and holds a subject steady across up to five reference images.

- **Model ID:** `spacexai/grok-imagine-image-2.0`
- **Type:** image
- **Providers:** xai
- **Context window:** Not applicable
- **Maximum output tokens:** 0
- **Pricing:** $0.06/image
- **Canonical page:** https://vercel.com/ai-gateway/models/grok-imagine-image-2.0

## Supported parameters

Detailed capability metadata has not been reported for this model.

## Example

```ts
import { experimental_generateImage as generateImage } from 'ai';

const result = await generateImage({
  model: 'spacexai/grok-imagine-image-2.0',
  prompt: 'A red balloon on a wooden table.'
});
```

## About

Grok Imagine Image 2.0 was released August 7, 2026 as SpaceXAI's text-to-image and image editing model. It generates images from a prompt and edits existing ones, and the emphasis across both is consistency: a subject, style, or product should survive an edit rather than drift into something new.

Typography is the clearest improvement. Grok Imagine Image 2.0 plans layout and small text the way a designer would, which makes it usable for posters, product shots, and interface mockups where legible text decides whether an image is usable at all.

Editing works at the region level. You point at an area and change only that area, select precise regions through segmentation, or remove a background to export a subject on transparency. Up to five reference images can be supplied at once to hold a character, style, or product consistent across a set. Smart Resize recomposes a finished image into other aspect ratios by filling the new frame rather than cropping it.

On Arena leaderboard standings at launch, Grok Imagine Image 2.0 placed second worldwide in both text-to-image and image editing. Leaderboard positions move week to week, so treat that as a snapshot.

A request can return up to 10 images, at 1K or 2K resolution. Generate images with `generateImage` from the AI SDK, and pass images in the prompt to edit them.

## What to consider

Edits bill for both the input image and the output image, so an iterative workflow that refines the same frame repeatedly costs more than the per-image rate suggests. Check the pricing panel on this page and budget for the number of passes your workflow actually takes.

Text rendering is strong in English and less reliable in Simplified Chinese, so verify output before shipping localised assets. Independent reviews also report that style transfer is less dependable than subject consistency: Grok Imagine Image 2.0 is a better choice when you need a character or product to persist across edits than when you need a specific artistic style reproduced faithfully.

Image models on AI Gateway are called through `generateImage` in the AI SDK. The Chat Completions, Responses, and Messages APIs do not serve image generation.

## When to use

### Best For

- **Typography-heavy images** such as posters, product shots, and mockups where small text has to stay legible
- **Region-level editing** that changes one part of a frame while the rest of the image holds
- **Consistent subjects** held across a set using up to five reference images
- **Aspect ratio variants** produced by Smart Resize recomposing rather than cropping
- **Transparent exports** from background removal, ready for compositing

### Consider Alternatives When

- **Faithful style transfer** is the goal, since style reproduction is less reliable than subject consistency
- **Simplified Chinese text** appears in the image, where text compliance is weaker than in English
- **Video output** is needed, where Grok Imagine Video generates moving footage rather than stills
- **Single cheap generations** are enough, where Grok Imagine Image costs less without editing or references

## Best for

- **Typography-Heavy Images:** Posters, product shots, and mockups where small text has to stay legible
- **Region-Level Editing:** Changing one part of a frame while the rest of the image holds
- **Consistent Subjects:** Up to five reference images keeping a character or product stable across a set
- **Aspect Ratio Variants:** Smart Resize recomposing a finished image instead of cropping it
- **Transparent Exports:** Background removal producing subjects ready for compositing

## Consider alternatives

- **Faithful Style Transfer:** Reviews report style reproduction is less reliable than subject consistency
- **Simplified Chinese Text:** Text compliance is weaker outside English
- **Video Output:** Grok Imagine Video generates moving footage rather than stills
- **Single Cheap Generations:** Grok Imagine Image costs less when you do not need editing or references

## Frequently asked questions

### What is Grok Imagine Image 2.0 built for?

Text-to-image generation and image editing, with an emphasis on keeping a subject, style, or product consistent across edits.

### How do I call Grok Imagine Image 2.0 on AI Gateway?

Use `generateImage` from the AI SDK with `spacexai/grok-imagine-image-2.0`. To edit an existing image, pass images in the prompt. The Chat Completions, Responses, and Messages APIs do not serve image generation.

### How many reference images can Grok Imagine Image 2.0 take?

Up to five at once. Use them to hold a character, style, or product consistent across a set of generations.

### How many images can one request return?

Up to 10, at 1K or 2K resolution.

### How does Grok Imagine Image 2.0 price image edits?

An edit bills for both the input image and the output image. See the pricing panel on this page for current rates, and budget for the number of passes an iterative workflow takes.

### Can Grok Imagine Image 2.0 render text inside an image?

Yes, and it plans layout and small text deliberately, which makes it suitable for posters and mockups. Text compliance is weaker in Simplified Chinese than in English, so verify localised assets before shipping.

### Can Grok Imagine Image 2.0 change one part of an image?

Yes. Point at a region to edit only that area, use segmentation to select precise areas, or remove the background to export the subject on transparency.

### Does Grok Imagine Image 2.0 support Zero Data Retention?

Yes, Zero Data Retention is available for this model. Zero Data Retention is offered on a per-provider basis. See https://vercel.com/docs/ai-gateway/capabilities/zdr for details.

## Links

- [Model page](https://vercel.com/ai-gateway/models/grok-imagine-image-2.0)
- [AI Gateway documentation](https://vercel.com/docs/ai-gateway)
- [Provider model documentation](https://docs.x.ai/developers/models)


If you encounter incorrect information, can’t find what you need, need a capability Vercel doesn’t currently offer, or have any other feedback to report, please do it by following the instructions at https://vercel.com/feedback.md.