✦ CHATGPT GUIDE · 7 MIN READ

Make a GIF with ChatGPT Images 2.6

A practical way to request a looping GIF in ChatGPT, the Flare and Sunburst limits OpenAI actually publishes, and how that compares with a one-click photo GIF.

Quick answer

In ChatGPT, ask for a looping GIF and describe what should move, or upload a still and ask ChatGPT to animate it. GIF creation is still rolling out. Frame count, frame timing, and the Flare and Sunburst model names are documented for the API, not as buttons in chat. For a selfie that should stay the same photo, use the photo-to-GIF maker and shrink the file with the GIF Compressor.

1. How to make a GIF in ChatGPT

ChatGPT Images 2.6 is the image experience inside ChatGPT on the web, iOS, and Android, on every subscription tier. Flare and Sunburst are the API model names, gpt-image-2.6-flare and gpt-image-2.6-sunburst. The ChatGPT help article does not ask you to choose either name in the app.

  1. Open ChatGPT and start an image. Ask in the conversation, or select More and then Images. Image generation may take a few minutes, depending on complexity, and you can keep using ChatGPT while an image is being created. The help article does not publish a separate timing for GIFs.
  2. Ask for a looping GIF and name the motion. OpenAI’s own example is: “Make a looping GIF of a paper boat bobbing on the water.”
  3. Or animate a still you already have. Upload the image and ask ChatGPT to animate it. You can also save a photo as a reference and later type @ plus the name you gave it. That reference flow is documented for images; the help article does not say it changes GIF length or frame settings.
  4. Revise only a GIF ChatGPT created. Describe the change in the conversation. Open that GIF to adjust playback speed, save it, or share it.
  5. Do not expect ChatGPT to edit a GIF you upload. The help article says ChatGPT can’t edit an animated GIF you upload. For a file you already have, use the GIF Compressor, GIF Resizer, or GIF Speed Changer.

For a selfie, one small action is easier to follow than a scene change. This prompt is a practical example in the same shape as OpenAI’s guidance (name the motion, return to the start, say whether the camera moves). It is not a sample output from ChatGPT.

SELFIE PROMPT TO TRY

“Animate this uploaded selfie as a looping GIF. The person blinks once, then returns to the same expression and pose. Keep the camera still and the background unchanged. No text.”

If GIF creation is not available in your account yet, that matches the help article: it is rolling out, not described as present for everyone.

2. Parameters and limits, from the official docs

ChatGPT and the API do not expose the same controls. The table uses only what OpenAI’s help center and image guides publish. Where a chat control is missing, that means it is not documented — not that a hidden setting exists.

LimitChatGPTAPI (Flare and Sunburst)
Where it appliesChatGPT chatImage API
Models you nameNot listed. The product is ChatGPT Images 2.6.gpt-image-2.6-flare or gpt-image-2.6-sunburst
How you request a GIFAsk for a looping GIF, or upload a still and ask to animate it.POST /v1/images/generations with output_format set to gif
Frame countNot published as a chat control.frame_count: 2–64. Default 9.
Frame timingOpen the GIF ChatGPT made to adjust playback speed. No millisecond control is documented.frame_duration_ms: 20–2000 ms per frame, rounded to 10 ms. Default 200.
Frame sizeAn aspect-ratio picker exists for images. The GIF section does not say it sets GIF frame size.size: auto, or WIDTHxHEIGHT with edges that are multiples of 16. Default auto.
EditsDescribe a change to a GIF ChatGPT created. An animated GIF you upload cannot be edited.The image edit endpoint does not support GIF output.
Other GIF restrictionsGIF creation is rolling out, so it may be missing on some accounts.No streaming, partial images, output_compression, or prompt_rewriter. frame_count and frame_duration_ms are GIF-only.

On the API, both Flare and Sunburst can generate a GIF. OpenAI describes Flare as the faster model for everyday iteration, and Sunburst as the model for more detail and more consistent edits. The announcement says Flare can generate images in as little as six seconds when quality is set to low. That figure is about images, not a published GIF timing. The GIF guide’s worked example uses Sunburst at 512x512, quality high, the default complexity high, frame_count 9, and frame_duration_ms 200, which OpenAI states is a 1.8-second loop.

size is the size of each finished frame. Frame count, frame dimensions, and the number of images (n) have to fit the model’s size limits and the request’s pixel budget. The GIF section does not print a single pixel cap. If a request is too large, OpenAI says to lower frame_count, the frame dimensions, or n, or to leave size on auto.

Quality (auto, low, medium, high, xhigh, max) and complexity (low, medium, high) are general GPT Image 2.6 settings. The GIF example sets quality. The ChatGPT help article does not list them as GIF controls in chat.

Two easy mix-ups are worth separating. First, the API edit endpoint does not accept GIF output, while ChatGPT’s help article does say you can upload a still image and ask for an animation. Those are different surfaces. Second, transparent backgrounds are documented for PNG, WebP, and SVG. The guides do not say a GIF keeps an alpha channel, so this article does not claim transparent GIF stickers from output_format: gif.

Prompts and generated images are filtered under OpenAI’s content policy. The guides do not list a separate selfie rule.

3. The same selfie, two different GIFs

Imagine one front-facing selfie and two exports. This repository has original preset GIFs in /samples, and none of them is a selfie or a ChatGPT capture, so the cards below are placeholders in the prompt-gallery style. They are not screenshots, and they are not model output.

Placeholder · ChatGPT Images 2.6

Native GIF

The same selfie goes in as a still or a saved reference. ChatGPT generates a new loop from the prompt. Treat the face as generated, not as a locked photo.

Placeholder · AI GIF Studio

One-click photo GIF

The same file goes into the photo-to-GIF maker. Float, pulse, or pan moves the original pixels. The tool does not invent a blink or a new expression.

FeatureChatGPT native GIFAI GIF Studio one-click
Same selfieUpload the still, or refer to a saved photo, and ask for a loop. ChatGPT generates a new animation.Upload the same JPG, PNG, or WebP. The file’s own pixels are what move.
Kind of motionWhatever you describe, such as a blink that returns to the first pose. The result is generated, so the face can change.A preset camera move: horizontal float, pulse zoom, or vertical pan. Faces and objects are not redrawn.
Length and sizeChatGPT does not publish frame count or pixel size for this. API defaults are 9 frames and 200 ms each (1.8 seconds in OpenAI’s example).16 frames, about 1.6 seconds, exported at 480×320.
After you have the fileSave or share from ChatGPT. Playback speed is adjusted on a GIF ChatGPT created.Download in the browser, then compress, resize, or change speed.
PrivacyThe photo is sent to ChatGPT.The photo is processed in your browser. AI GIF Studio does not upload it.

The studio side is the photo-to-GIF maker: JPG, PNG, or WebP under 8 MB, fitted to a 480×320 frame, 16 frames, processed locally. It is a camera-style effect, not ChatGPT Images and not a generative scene. If the ChatGPT download is heavy, put that GIF through the GIF Compressor, which can reduce dimensions, colors, and frame density in the browser.

4. Which one to open

  • The photo should stay the photo. Use the one-click maker, then compress it if you need a smaller file.
  • The motion has to be invented. Use ChatGPT Images 2.6 and describe a short action that returns to the first frame. Expect a generated result, and expect GIF support to be missing until rollout reaches your account.
  • You want a meme or sticker that is not your face. Open the meme GIF templates to edit a caption and download a Canvas sticker of an original mascot. Prompt ideas still live in the reaction GIF prompts and the GIF Prompt Generator. Those pages are preset drawings or prompt text, not Images 2.6 output.

For a longer prompt checklist — subject, one action, style, camera, and a matched first and last frame — see How to Make an AI GIF.

Try it on your photo

Make the one-click GIF

Upload a selfie for a private looping GIF, or start from an original meme template and edit the caption. These pages export preset browser GIFs. They do not call ChatGPT.

Try photo to GIF Try meme GIF templates

Sources: Images in ChatGPT · Introducing Images 2.6 · Image generation · Image prompting. Product behavior can change while GIF creation is rolling out.