New

GPT Image 2.6 is now live

Try GPT Image 2.6
LogoAI Image 2
Sign In
AI Image 2

AI Image 2

Type a prompt, pick a ratio, and generate posters, product photos, and illustrations with readable in-image text.

Cost
10 credits / image
Ratios
11 aspect ratios
Max size
4K
API
Generate + status

GPT Image 2 examples from the prompt library

Published entries from our prompt library. Open a card in the workspace and adapt it.

These images come from our prompt library and were not all made on this site.

What is AI Image 2?

AI Image 2 is a browser workspace for the GPT Image 2 model family. Describe an image in plain words or upload a reference photo, choose a model and aspect ratio, and generate. AI Image 2.0 renders legible text inside images, so posters, infographics, and product shots come out ready to use.

Every AI Image 2.0 image costs 10 credits regardless of size. The same workspace also edits existing photos: upload a reference, describe the change, and the model keeps the subject while it restyles the scene. A free prompt gallery and a developer API cover the same generation contract.

AI Image 2 is an independent service and is not affiliated with, endorsed by, or sponsored by OpenAI.

Input
Text prompt, or text + reference images
Output
Image file, downloadable from the result
Aspect ratios
auto + 10 more
Resolutions
1K, 2K, 4K
Credits per image
10 (AI Image 2.0)
Free credits
Offered to new accounts
Models
60+ in the picker
API
POST /api/generate, GET /api/status
Commercial use
Subject to terms

How it works

The composer above hands your prompt and settings to the workspace.

  1. Step 1

    Write a prompt

    Describe the image in plain words. Include the exact text you want rendered — a poster title, a label, or a short slogan.

  2. Step 2

    Pick a ratio

    Choose one of the supported aspect ratios in the composer. Square and portrait work well for social posts; 16:9 and 21:9 fit banners.

  3. Step 3

    Add a reference if you have one

    Attach a photo to edit instead of starting from scratch. The model keeps the subject and changes what you describe.

  4. Step 4

    Generate

    Sign in, confirm the credit cost, and submit. Results land in your workspace history for download.

What people make with GPT Image 2

Five jobs the model handles well.

Luxury handbag in a complete fashion campaign scene with premium textures

Product photography without a studio

Wrap a product in a full campaign scene — lighting, background, props — while keeping the item itself accurate. The example stages a handbag in a complete fashion spread.

Open the workspace
Commercial storyboard of a cold-pressed juice bottle with splashes and droplets

Ad creative and commercial posters

Condensation on a bottle, a cap mid-open, a splash frozen in place. The model renders the small physical details that sell a drink or a snack.

Open the workspace
Vintage travel magazine poster with bold typography and halftone textures

Posters with real typography

Headlines, subheads, and captions stay legible at print sizes. Mid-century halftone, bold stencil, or clean editorial grids all hold up.

Browse the prompt gallery
Photo paired with handmade paper texture and wax-crayon sketch in muted tones

Reference-based photo editing

Upload a picture and describe the change. Swap a background, restyle a product shot, or merge a photo with a sketch without losing the subject.

Open the workspace
Hand-drawn sketchbook page with ink landmarks and handwritten labels

Illustration and editorial art

Loose ink sketchbook pages, layered paper cutouts, and hand-drawn journals come out with the imperfect texture of real media.

See more examples

What the generator gives you

The same controls appear in the composer and in the full workspace.

Typographic cityscape where New York landmarks form bold geometric letterforms

Readable in-image text

Put the exact words in the prompt — a poster headline, a menu, a label — and the letters come out correctly spelled and positioned.

Photo paired with handmade paper texture and wax-crayon sketch in muted tones

Reference editing

Pass one or more photos with the prompt to restyle, recompose, or extend them while preserving the subject.

Hyper-realistic 3D isometric travel poster of a detailed map with landmarks

Structured layouts

Multi-panel compositions — storyboards, infographics, isometric maps — keep their panels ordered and labeled.

1K to 4K output

Pick 1K, 2K, or 4K in the workspace. Eleven aspect ratios are supported; 5:4 and 4:5 render at 1K, and 1:1 skips 4K.

60+ image models in one picker

Switch between AI Image 2.0, GPT Image 2.5, NanoBanana, Seedream, Flux, and more without leaving the page.

Developer API

POST /api/generate returns a task_id; poll /api/status until SUCCESS. Full parameters are on the docs page.

Models and specs

The three GPT Image 2 family models and what each accepts.

ModelInputAspect ratiosResolutionsCredits / image
AI Image 2.0Text, or text + reference imagesauto + 101K, 2K, 4K10 flat
GPT Image 2.5Text, or text + up to 16 imagesauto + 121K, 2K, 4K4 / 6 / 10 by size
AI Image 1.5Text, or text + image editSquare, portrait, landscape1024px output10 flat

How AI Image 2.0 compares

Credit costs and strengths of the other models in the same picker.

ModelCredits / imageBest for
AI Image 2.010 flatPosters, infographics, UI text, reference edits
NanoBanana 2.14–9 by sizePhotoreal scenes, extreme ratios to 1:8
Seedream 5 Pro9Artistic and editorial styles
Flux 310Clean product and concept renders
Z-Image2Fast drafts at the lowest cost

Read the full NanoBanana 2 vs GPT Image 2 comparison

Credits, with or without a subscription

One-time packs and monthly plans share the same per-image rates: 10 credits for AI Image 2.0, 4–10 for GPT Image 2.5 by resolution. Failed generations are not charged.

See pricing

FAQ

What can I make with AI Image 2?

You can create images from text and edit uploaded photos. Make product visuals, posters, illustrations, avatars, and images for social posts.

Can I edit my own photo?

Yes. Upload it in the workspace and describe what you want to change, such as the background, colors, or style.

Can I add words to an image?

Yes. Include the exact words in your prompt, such as a poster title or a short slogan, and check the spelling in the finished image.

How much does an image cost?

An AI Image 2.0 image costs 10 credits at any size. GPT Image 2.5 costs 4, 6, or 10 credits depending on resolution. See Pricing for current packages.

Can I use images for my business?

Commercial use is subject to our service terms. Make sure you have permission to use any uploaded photos, logos, or likenesses.

Is AI Image 2 an OpenAI website?

No. AI Image 2 is an independent service and is not affiliated with, endorsed by, or sponsored by OpenAI.

Which aspect ratios are supported?

Eleven values: auto, 1:1, 5:4, 9:16, 21:9, 16:9, 4:3, 3:2, 4:5, 3:4, and 2:3.

What resolution can I export?

1K, 2K, or 4K. 4K is unavailable at 1:1, and 5:4 / 4:5 only render at 1K.

Is there a developer API?

Yes. Create an API key under Dashboard > API Keys, POST to /api/generate, and poll /api/status. Full documentation lives at /docs.

What else is in the dashboard?

AI video generation, AI speech, and a full set of image, video, and audio tools — background removal, upscaling, photo translation, and more.

What do you want to create?

Start with your idea in the composer above.

Back to the composer

AI Image 2 is an independent service and is not affiliated with, endorsed by, or sponsored by OpenAI.