Two versions, one place
Grok Image costs 3 credits, good for trying ideas. Grok Image 2.0 adds detail at 5 credits (low) or 8 (medium).
xAI · Image model
xAI's Grok image models, both versions in one place: generate from a prompt or edit with reference images, no X Premium needed.
Both of xAI's Grok image models are here: Grok Image is quick and cheap, and Grok Image 2.0 adds finer detail with low or medium quality. Generate from a prompt, or edit with up to 5 reference images.
Opens the generator with the model selected. No X Premium needed, and failed runs are refunded. This site uses a third-party API and isn't affiliated with xAI.
Neon billboard in the rain, with clear text
Each image is shown with the prompt that made it. Like one? Click "Use this prompt" and make your own.
Grok Image · 1:1
A vibrant street food scene in Bangkok: a vendor preparing pad thai in a flaming wok, golden noodles mid-toss, colorful condiments, busy market background with warm tungsten lighting, food photography
Grok Image costs 3 credits, good for trying ideas. Grok Image 2.0 adds detail at 5 credits (low) or 8 (medium).
Add up to 5 reference images and say what changes and what stays: new background, new style, same character in a new scene. No masks.
Product shots, portraits, landscapes, watercolor, anime and 3D cartoons all work. Just name the style in your prompt.
Short lines like poster titles and thumbnail text come out readable when you put them in quotes.
An image here usually takes 15 to 25 seconds, a bit longer when it's busy.
No X Premium or SuperGrok subscription. Each image uses credits, and failed runs are refunded automatically.
| Option | Grok Image | Grok Image 2.0 |
|---|---|---|
| Mode | Text to image, edits with references | Same |
| Resolution | 1K (about 1 MP) | Same |
| Aspect ratio | 1:1, 3:4, 2:3, 9:16, 4:3, 3:2, 16:9 | Same |
| Quality | Single | Low, Medium |
| Reference images | Up to 5, max 10MB each, JPG, PNG or WebP | Same |
| Images per run | 1 | 1 |
| Prompt | Up to 1,000 characters | Same |
| Credits per image | 3, plus 1 per reference image | Low 5, medium 8, plus 1 per reference image |
1K and one image per run are the limits of the API this site uses; xAI's own API also offers 2K.
Here are example images made with Grok. Click any image to zoom in.
Food, architecture, pets and landscapes with light and texture that look photographed.
How to start: Describe the subject, setting and light, then name the kind of photo, like "food photography."
A vibrant street food scene in Bangkok: a vendor preparing pad thai in a flaming wok, golden noodles mid-toss, colorful condiments, busy market background with warm tungsten lighting, food photography
Need many product shots? Try batch generation.
Use this promptUpload an image and say how to change it, like making a game scene more realistic.
How to start: Upload the image and write "Keep … the same, change … to …."
Make this scene more realistic but still keep the game vibes
Short lines like titles and slogans go right into the image.
How to start: Put the text in quotes; shorter lines come out cleaner.
A neon-lit cyberpunk figure standing in the rain beneath a holographic billboard, cinematic lighting
Watercolor, blind-box toys or 3D cartoons, just name the style.
How to start: Try a few with Grok Image, then make the final with 2.0.
3D character design, blind box toy style. A cute chibi girl wearing a fluffy dinosaur hoodie and oversized sneakers. She is sitting on a giant macaron. Soft clay texture, matte finish. Clean pastel color palette: mint green, baby pink, and cream. Studio lighting, soft shadows, 3D render, C4D, Octane render, high fidelity, 4k.
Scenes and characters for games or stories, with plenty of atmosphere.
How to start: Describe the scene, light and mood, then add a medium like "oil painting texture."
Dark fantasy concept art. A fallen knight in rusted, ornate armor kneeling in a ruined gothic cathedral. A massive, glowing red rune floats in the air above him. Fog rolls across the floor. Dramatic, high-contrast lighting. Shadows hide terrifying creatures in the corners. Oil painting texture, intricate details, cinematic composition, gloomy atmosphere.
16:9 for video covers, 9:16 for phone screens.
How to start: Pick the ratio first and keep the subject near the center.
A golden retriever puppy playing in autumn leaves
Pick a model: start with Grok Image to explore, then switch to Grok Image 2.0 (Low or Medium) for detail.
Write a prompt or add references: describe the subject, setting, light and style. To edit, add up to 5 images and say what changes and what stays.
Pick a ratio and generate: the button shows the credits first, and you can zoom in and download the result.
Write subject, setting, light and style in one sentence.
Put any text in quotes; shorter is more reliable.
With references, start with "Keep … the same," then say what to change.
Prompts can be up to 1,000 characters; cut repeated adjectives if you run out.
| Feature | Grok Image | Grok Image 2.0 | MiniMax H3 Image | Nano Banana 2.1 | Seedream 5.5 |
|---|---|---|---|---|---|
| Max resolution | 1K | 1K | 4K | 4K | 2K |
| References | Up to 5 | Up to 5 | Up to 16 | Up to 1 | Up to 10 |
| Per run | 1 | 1 | 1–10 | 1–4 | 1 |
| Quality | Single | Low, Medium | — | — | — |
| Best at | Cheap tries to find a direction | Detail, product shots, edits | 4K, multi-image edits | Several images per run | Multi-round edits, transparency |
| Credits | 3 | From 5 | From 3 | From 2 | From 3 |
All of these are in the same generator. Not sure? Run one prompt with Grok Image first, then try the same prompt on another model.
Pay per image with credits, no subscription: Grok Image is 3 credits, Grok Image 2.0 is 5 (Low) or 8 (Medium), plus 1 per reference image. The button shows the exact amount before you generate. Packs start at ¥9.9 and never expire; failed runs are refunded automatically.
Using Grok's image features on X means an X Premium or SuperGrok subscription, and xAI's API needs your own key and billing. Here you open the page and pay per image.
See credit packsGrok is xAI's AI assistant, and Grok Imagine is its image and video feature. xAI's API offers two image models, which our API provider calls Grok Image and Grok Image 2.0. This site uses them through a third-party API and isn't affiliated with xAI.
Grok Image costs 3 credits and is good for quick tries. Grok Image 2.0 adds detail and offers Low and Medium quality, with Medium the sharpest. Ratios, references and resolution are the same.
No. You don't need an X account or a SuperGrok plan. Sign in here and pay per image with credits.
Add 1 to 5 reference images in the generator, up to 10MB each, as JPG, PNG or WebP. Start your prompt with "Keep … the same," then say what to change.
Both versions output 1K here, about 1 megapixel: 960×960 at 1:1 or 1280×720 at 16:9. xAI's own API also offers 2K; our API provider currently offers 1K only.
Your first one can be: new accounts get 4 credits, enough for one Grok Image. After that you pay per image, with packs from ¥9.9 and no subscription.
Grok Image is 3; Grok Image 2.0 is 5 (Low) or 8 (Medium); each reference adds 1. The button shows the exact amount.
This site and our API provider both moderate content, and explicit, violent or impersonating requests are blocked. A blocked run doesn't use credits; try rewording it.
Failed runs are refunded to your account automatically. Our API provider is sometimes busy; wait a moment and try again.
No key is needed here; just sign in and generate. If you want the API itself, xAI offers it on its developer platform.
Under the terms of xAI and our API provider, the images you generate belong to you. Commercial use must follow their usage rules, such as not passing images off as human-made or removing provenance marks. Make sure you have rights to your prompts and reference images, you're responsible for content involving other people's likeness or trademarks, and follow the laws where you live.
Grok makes one image per run here; generate again to compare more. This site makes images only, not video.
Try a few with Grok Image to find the look, then make the final with Grok Image 2.0.
From 3 credits per image. Failed runs are refunded.