# Grok Imagine Image 2.0: xAI's image model — typography, in-place edits and 5 references

> Grok Imagine Image is a family of image models from xAI, the company behind Grok. The current Imagine Image 2.0 came out on August 7, 2026: it follows instructions closely, keeps small text sharp, edits a selected area, removes backgrounds and takes up to 5 references. It ranks 6th of 82 on Arena and costs $0.04 per image in the API. Grok Imagine Image is available in the AI2Face generator as Grok Image 2.

Developer: xAI · Released: 2026-08-07 · Updated 2026-10-28
URL: https://ai2face.com/models/grok-imagine-image

## Key facts

| | |
|---|---|
| Developer | xAI |
| Release date | Image 2.0 — August 7, 2026; the first model, Aurora — December 2024 |
| API models | grok-imagine-image-2.0, grok-imagine-image, grok-imagine-image-quality |
| Editing | Magic wand, segmentation, background removal, Smart resize |
| References | Up to 5 images per generation |
| Per request | Up to 10 images |
| API price | 2.0 — $0.04; base — $0.02; quality — $0.05 per image |
| Arena rank | 6th of 82 in generation (Oct 26, 2026) |
| In AI2Face | Yes, in the generator — Grok Image 2 |

## What is Grok Imagine Image

Grok Imagine Image is a family of image generation models from xAI. xAI released its first image model, Aurora, in December 2024. Grok Imagine then added Imagine Image and its upgraded Pro version, and on August 7, 2026, **Imagine Image 2.0** arrived.

xAI puts the goal of 2.0 this way: images you can use in real work. In the Grok apps and on grok.com/imagine it became Quality mode; in the API it's `grok-imagine-image-2.0`. An overview of all of Grok Imagine is on the [Grok Imagine](/models/grok-imagine) page, and video is on the [Grok Imagine Video 1.5](/models/grok-imagine-video-1-5) page.

On the Arena leaderboard as of October 26, 2026, the newer Image 2.0 build ranks 6th of 82 and the August build 8th. The earlier Imagine Image ranks 25th, and Pro 29th. At launch, xAI reported second place in both generation and editing.

## What Grok Imagine Image 2.0 can do

**Typography and layout.** The model plans fonts and placement like a designer and keeps small text sharp even in dense, multi-part layouts.

![A HARBOR JAZZ WEEK festival poster in mid-century style: dates, a list of bands and the venue, saxophones and waves](https://ai2face.com/_astro/poster.Dl8DAvKa_1wLeln.jpg "A poster with six lines of text. The text is sharp, but the model repeated one line of the list — check your lists. Grok Image 2 in AI2Face")

**In-place editing.** The magic wand changes only the area you point at. Segmentation selects precise zones — just a sleeve or just a mug.

**Background removal.** A subject is exported on a transparent background — ready for a product card.

![A handmade speckled green ceramic mug on a white background with a soft studio shadow](https://ai2face.com/_astro/product.B4wEpRzt_ZBGjHe.jpg "A product shot on white — a source for cutting out and for the E-Commerce Photos template. Grok Image 2 in AI2Face")

**Up to 5 references.** One generation can mix up to 5 source images: combine characters, transfer a style, build a scene.

**Smart resize.** The model recomposes the frame for a new aspect ratio — from 1:2 and 9:16 to 16:9 and 2:1 — and fills in what's missing.

**15 templates.** Ready-made workflows: photo edit, product color change, editorial product poster, collage, mascot, background removal and replacement, e-commerce photos, UGC shots, professional headshots, icons, character sprites, a UI kit, emoji and merch.

![A mascot sheet: a cartoon fox in a green scarf in four poses — waving, thumbs up, holding coffee, jumping](https://ai2face.com/_astro/mascot.BKxtaYXz_qPGf5.jpg "One character in four poses — what the Mascot Maker template does. Grok Image 2 in AI2Face")

**A world for video.** You can generate a character, locations and props separately in one consistent style, then build a clip from them in [Grok Imagine Video 1.5](/models/grok-imagine-video-1-5).

## Grok Imagine Image versions and pricing

| | Imagine Image | Imagine Image quality | Imagine Image 2.0 |
|---|---|---|---|
| API ID | grok-imagine-image | grok-imagine-image-quality | grok-imagine-image-2.0 |
| Price per image | $0.02 | $0.05 | $0.04 |
| Arena rank (Oct 26, 2026) | 25 | 29 (Pro) | 6 and 8 |
| Status | earlier | earlier | recommended by xAI |

xAI charges a flat price per image regardless of prompt length. For edits, both the input image and the result are billed. One request can return up to 10 images.

## What Grok Imagine Image is good for

- **Posters, covers and infographics** with small text.
- **Product cards** — background removal, product color changes, e-commerce photos.
- **Mascots, icons, emoji and stickers** from ready-made templates.
- **In-place photo edits** — replace an object without touching the rest.
- **Resizing for social** — one frame in 9:16, 1:1 and 16:9.
- **Preparing characters and locations** for video.

How to get images in Grok and what's free is covered in the [Grok Image](/blog/grok-image) guide.

## Grok Imagine Image vs alternatives

| | Imagine Image 2.0 | [GPT Image 2.5](/models/gpt-image-2-5) | [Nano Banana 2](/models/nano-banana-2) | [Muse Image](/models/muse-image) |
|---|---|---|---|---|
| Developer | xAI | OpenAI | Google DeepMind | Meta |
| Arena rank (Oct 26, 2026) | 6 | 1 (Sunburst), 2 (Flare) | 11 | 9 |
| Editing | region, segmentation | mask and prompt | prompt | up to 10 images |
| Background removal | yes | transparent background | — | no |
| References | up to 5 | several, limit not disclosed | up to 14 | up to 10 |
| API price | $0.04 | per token | from $0.045 | $0.01 |
| In AI2Face | yes | yes | yes | yes |

Image 2.0 sits in Arena's top 10 alongside models from OpenAI, Microsoft and Google, and it's more convenient than them for in-place editing. GPT Image 2.5 is ahead on quality, Nano Banana 2 takes more references, and Muse Image is the cheapest.

## How to write prompts for Grok Imagine Image

- **Put text in quotes or after cues like "title," "below it,"** and end with "no other text."
- **Keep lists to short lines** and check the result: a line got duplicated in our poster.
- **For edits, select the area** and describe only the change: "replace the mug with a glass one, leave everything else."
- **For products,** ask for a "pure white seamless background" — it makes cutting out easier.
- **For a series of one character,** list the poses separated by commas in one prompt.

The prompt for the poster above:

```
A jazz festival poster in mid-century modern style: title HARBOR JAZZ WEEK at the top
in large letters, below it OCTOBER 10-16, then a list of four small lines: Mira Quartet,
The Low Tides, Sam Ortega Trio, Blue Lantern Band, and at the bottom Pier 9 Stage.
Abstract saxophone and waves shapes, navy, mustard and coral palette,
all text exactly as written and sharp, no other text, no logos
```

## How to use Grok Imagine Image in AI2Face

In the AI2Face generator, the model is called **Grok Image 2**.

1. Sign up — new accounts get starter credits.
2. Open the generator and pick Grok Image 2.
3. Write a prompt, choose the aspect ratio and 1K or 2K, and add up to 5 references if needed.
4. The credit cost is shown before you start. Credits don't expire, and you don't need a SuperGrok subscription.

[GPT Image 2.5](/models/gpt-image-2-5) and [Nano Banana Pro](/models/nano-banana-pro) are right next to it: run the same prompt on all three and keep the best result.

## Limitations

- **Lists and repeats.** Long lists sometimes get a duplicated or missing line.
- **Background text.** The model may fill background signage with gibberish.
- **Up to 5 references** — fewer than [Nano Banana](/models/nano-banana) and Muse Image.
- **Resolution** — 1K or 2K, no 4K.
- **xAI moderation** — some requests are rejected.

## Bottom line

Grok Imagine Image 2.0 is a model for practical work: posters with small text, product shots with the background removed, mascots and in-place edits — 6th on Arena at $0.04 per image. Use it when you need to fix part of an image or quickly make a series in one style. For the most accurate text, use [GPT Image 2.5](/models/gpt-image-2-5); for photos of people from several references, [Nano Banana Pro](/models/nano-banana-pro). All three are in the AI2Face generator.

## FAQ

### What can Grok Imagine Image 2.0 do?

It generates images with precise typography and layout, edits only a selected area, cuts a subject out onto a transparent background, recomposes a frame for a new aspect ratio and builds a scene from 5 source photos. In the Grok apps it's Quality mode.

### How much does Grok Imagine Image cost?

In the xAI API, Image 2.0 costs $0.04 per image, the earlier Imagine Image $0.02, and the quality version $0.05. The price doesn't depend on prompt length; for edits, both the input and the new image are billed.

### How do you edit photos in Grok Imagine?

Point at an area with the magic wand and describe the change — the model leaves everything else alone. Segmentation handles precise areas, and background removal exports products on a transparent background.

### How is Image 2.0 different from the earlier Grok Imagine Image?

2.0 follows instructions more closely, keeps small text and layout intact, and preserves the source during edits. On Arena the gap is large: 6th place versus 25th for the earlier model.

### Is Grok Imagine Image available in AI2Face?

Yes. In the AI2Face generator it's called Grok Image 2, and no SuperGrok subscription is needed. Every image on this page was made with it.

## Sources

- https://x.ai/news/grok-imagine-image-2
- https://docs.x.ai/developers/models
- https://docs.x.ai/developers/model-capabilities/imagine
- https://x.ai/news/grok-image-generation-release
- https://arena.ai/leaderboard/text-to-image