ChatGPT Images: How to Create and Edit Pictures

Create a picture, change part of a photo, or animate a still in ChatGPT Images. See which plans include thinking and what to do when a limit appears.

Published

ChatGPT Images: How to Create and Edit Pictures

ChatGPT Images is OpenAI’s tool for creating and editing pictures inside ChatGPT. It runs on Images 2.6, and you can use it on the Free plan. You can upload a photo and change one detail without having to choose between Sunburst and Flare.

Can ChatGPT generate images?

Yes. Describe what you want in a conversation, from a shop-window poster with a headline to an illustration with a transparent background. If you already have the right subject, start with its photo. You can change the background, add lettering or adjust the shape of the picture for the space where it will appear.

OpenAI gives a good example of how little you need to bring. Its Images 2.6 announcement shows an owl drawn in pen on lined paper, followed by an animated version of that owl. The sketch supplies the character. The next instruction supplies the movement. You can use the same approach when the curve of a hat brim or the shape of a mascot’s ears matters more than your ability to describe it. See OpenAI’s owl example and Images 2.6 announcement.

OpenAI’s Images 2.6 announcement with examples of generated imagery and animation
ChatGPT Images

Where it runs. Use ChatGPT Images on the web, iOS or Android. OpenAI also offers generation and editing in Codex, its coding tool, and through the API, which lets developers build it into their own software.

Which model to pick. In ChatGPT, just ask for the image. OpenAI documents a choice between GPT-Image-2.6 Flare and GPT-Image-2.6 Sunburst for developers using the API. Flare favors quick drafts at low quality. Sunburst favors fine detail and consistency across edits at medium or high quality. If you’re building a product and need to choose, see Sunburst and Flare on four design briefs.

Which plans include ChatGPT Images?

Every tier includes ordinary image creation. Free costs $0 per month and includes limited image creation. Go adds more creation, Plus supports more complex image requests, and Pro lists faster, unlimited image creation subject to abuse guardrails. Business, Enterprise and Edu also have the base feature. Compare OpenAI’s current plan descriptions.

Images with thinking gives ChatGPT time to plan a picture before it draws. Consider it for a poster where a headline, a curved brim and a shop address all need to fit. As checked on November 22, 2026, OpenAI’s current Images FAQ lists thinking for Plus, Pro and Business, with Enterprise and Edu still to come. Free and Go users can make the poster with ordinary generation. OpenAI’s Images FAQ.

How many images can I generate with ChatGPT Free?

OpenAI does not publish a fixed daily image count for Free in its current plan or help pages. Image creation can reach a limit separately from ordinary chat and from uploading files. If a notice appears, read which action it names. For an image-creation limit, wait until the reset time if one is shown, or try again later. While you wait, keep the original photo and write down the exact revision you want. OpenAI’s Free tier limit guidance.

For an occasional poster, start on Free. If limits regularly interrupt revisions, compare Go and Plus to decide whether you need more attempts or access to thinking. For sustained daily work, compare Plus and Pro against how often you have to stop.

How to create images with ChatGPT

Start from the material you want the final picture to resemble. A description is enough for an invented scene. Upload the original if a customer needs to recognize the actual product.

  1. Open Images

    Sign in to ChatGPT, or create a Free account. Ask for a picture in a new conversation, or open Explore and choose Images. For a layout to customize, open Templates under Images and pick a starting design. Templates are unavailable in Work mode.

  2. Give it a subject and a destination

    Describe what should appear and where the image will go. For a shop-window poster, specify a portrait layout and put the exact headline in quotation marks. You can give the width-to-height ratio in the prompt or use the aspect ratio picker.

  3. Open the result and revise it

    Select the picture. For a local change, choose Select, mark the area and describe the edit. On mobile, tap Next after highlighting. For a change that affects the whole picture, such as a different background, describe it directly in the conversation.

  4. Inspect and save

    Read every letter at the size your audience will see. Compare the product’s outline and shadow with the original. Open the finished picture and select Save to download it, Copy to use the clipboard, or Share to send it on. You can revisit generated pictures under Images.

Three prompts for ChatGPT Images

Try these on one milliner’s brief. Each gives you a different starting point and something specific to check.

Generate a poster. “Create a portrait shop-window poster for a spring hat fitting. Show a cream felt hat with a plum ribbon on a wooden stand. Set the headline ‘Spring Fitting’ above the crown. Leave the bottom quarter empty so I can add the address.” Check that the title is spelled correctly and the brim stays clear of the address space.

Edit a photo. “In the attached hat photo, change the plum ribbon to poppy red. Keep its width and knot, the cream felt, the angle of the brim and the shadow on the table.” This is the better starting point for a listing of a hat you sell. Compare the revised photo with the actual item before approving it.

A milliner’s hat annotated to distinguish the ribbon to change, the brim to keep and the shadow to inspect
A local edit has two parts. Mark what should change and name the details that must survive it.

Animate the approved still. “Make a looping GIF of this hat. A breeze lifts the loose ribbon end, then it settles into its starting position. Keep the hat, camera and light still.” A GIF is a short repeating animation. Watch the join where the last frame returns to the first. A jump there is easier to diagnose when only the ribbon moves.

Three clay flipbook pages show the ribbon resting, lifting and returning while the hat stays still
Give a loop an ending that can meet its beginning.

GIF creation is rolling out. Where it is available, you can animate an uploaded still, revise a GIF made in ChatGPT, change its playback speed, then save or share it. The editor cannot revise an already animated GIF you upload.

Can ChatGPT edit images without changing the rest?

A selection guides the edit like a pencil circle on a proof. OpenAI says changes can extend beyond the highlighted area. In the ribbon example, compare the brim edge, knot and shadow as well as the color. A shifted brim might be fine on a mood board. For a product listing, keep the original photo if the edit changes the hat you promise to deliver.

For another campaign featuring the same hat, ask ChatGPT to save its photo as a named reference. In a later chat, type @ followed by that name to bring it back. If the layout is easier to draw, use @, then Sketch on mobile, draw the arrangement and add your instructions. See how to use Sketch on a phone. When the same item needs to survive a whole catalog of edits, compare image editors for brand consistency.

How to delete images from ChatGPT

Save any image you want to keep on your device first. To remove a generated image from your ChatGPT collection, delete the conversation where you created it. On the web, use the menu beside the chat title. On mobile, press and hold the title to reach Delete. This removes the conversation along with the picture.

Which version of ChatGPT Images am I using?

Images 2.6 is the current release. Older dates explain why instructions may send you looking for a different model or menu. These milestones follow OpenAI’s dated ChatGPT release notes.

DateReleaseWhat you could do
December 16, 2025ChatGPT Images on web and mobileCreate or edit a picture and find it again in My images on web, iOS or Android.
April 21, 2026Images 2.0Generate across all plans. Images with thinking introduced planning before generation.
September 8, 2026Images 2.5Start from a template, draw a mobile sketch or share a prompt for someone else to adapt.
November 4, 2026Images 2.6Preserve more detail through revisions, reuse a named reference photo and, as access rolls out, make a looping GIF.

DALL·E and Sora. OpenAI retired the official DALL·E GPT from ChatGPT. If a tutorial sends you looking for DALL·E, open Images instead. Sora was OpenAI’s separate video product. Its web and app experiences have closed, and the looping GIF feature lives in ChatGPT Images. OpenAI’s Sora discontinuation notice.

Questions about using ChatGPT

What is the name of the ChatGPT image generator?

ChatGPT Images. You can ask for it directly in a conversation by describing the picture you want.

Is GPT-3 the same as ChatGPT?

No. GPT-3 is an older language model. ChatGPT is the product where you can chat, upload a photo and use Images 2.6.

What is an AI image?

It’s a picture a model generates or changes in response to instructions. A hat drawn from a description is generated. Recoloring the ribbon on an uploaded hat photo is an AI-assisted edit.

How to generate GPT?

If you mean a picture, ask ChatGPT to draw it. A custom GPT is a configured assistant, and personal accounts can no longer create new ones. OpenAI’s custom GPT availability guide.

Can I use Chat gbt for free?

Yes. A Free ChatGPT account includes image generation and editing. When your image allowance runs out, you can wait for it to reset.

Can someone tell if I'm using ChatGPT?

For a shared picture, OpenAI embeds origin information, including Content Credentials and an invisible watermark. A positive verification result indicates likely processing by an OpenAI tool. It does not identify the person who used it. OpenAI’s guide to verifying images.

Who can see my ChatGPT conversations?

Anyone with a conversation link shared from a personal account can see the material included in that link. Check the preview before sending a client an image from a chat that also contains drafts. OpenAI’s guide to shared conversations.

How do I know if someone else is using my ChatGPT?

On the web, open Settings, then Security and login, then Active sessions. Review devices you don’t recognize and log out of all sessions if you suspect someone else has access.

How can I see if someone used ChatGPT?

For past access to your own account, open Security history under Security and login. Review sign-in events and times. If you mean an image someone sent you, an origin check can identify signs of OpenAI processing without revealing the creator. OpenAI’s account activity instructions.

Where to go next

For your next shop update, try one photo you know well. Change a single detail, then compare the result at listing size. If the exact shape survives and the text reads cleanly, save the revision. If you want to explore a different art direction, look at alternatives to Midjourney. To choose a tool for your next brief, start with the best AI image generators for different kinds of work.

Get the newsletter for weekly field notes on using AI at work.

Published
Category
Design
Related Articles
Lovart: What It Makes and How to Start

Lovart: What It Makes and How to Start

See how Lovart turns a brief into campaign assets, how to start on its canvas, and which rights to check before delivering client work.Nov 22, 2026Design
Higgsfield GPU Harness: What It Is, What It Costs, How to Start

Higgsfield GPU Harness: What It Is, What It Costs, How to Start

Rent a cloud GPU from chat for Blender, Houdini or model training. See the card rates, the setup steps and the full rental cost of a Blender walk cycle.Nov 22, 2026Design
Google Flow: What It Is, What It Costs, and How to Start

Google Flow: What It Is, What It Costs, and How to Start

Make a first scene in Google Flow. See which plans include it, when to use a frame or an ingredient, and how Flow differs from Vids and Gemini.Nov 21, 2026Design
GPT Image 2.6 Sunburst vs Flare on Four Design Briefs

GPT Image 2.6 Sunburst vs Flare on Four Design Briefs

See eight first-try images, their prompts, timings, and costs across poster design, product photography, illustration, and storyboards.Nov 19, 2026Design
HeyGen Avatar 6 for Business Video

HeyGen Avatar 6 for Business Video

See how HeyGen Avatar 6 follows a script, where to select it in the app, and when training, sales and product videos still need original footage.Nov 19, 2026Design
LTX Desktop

LTX Desktop

Check whether your computer can run LTX Desktop, install the right release and turn one product photo into a clip you can edit locally.Nov 19, 2026Design
Kling Image 4.0 for Product Shots and Ad Campaigns

Kling Image 4.0 for Product Shots and Ad Campaigns

Use Kling Image 4.0 for product crops, editable poster copy and consistent campaign sets. See the supported ratios and what to check before approval.Nov 18, 2026Design
Looka AI Review

Looka AI Review

Looka AI reviewed for a small-business launch: logo workflow, Brand Kit, usable downloads, ownership boundaries and when to hire a designer.Nov 17, 2026Design
Newsletter

One letter, every Sunday.Working systems, not hot takes.

Weekly. No spam. Unsubscribe anytime.