Summary
- ChatGPT Images 2.6 creates animated GIFs from simple text prompts without requiring a starting photo.
- Two examples show a pelican at Lake Kerkini and an excited owl.
- OpenAI reports 50% faster image generation and greater editing consistency.
- Images 2.6 is available in ChatGPT, Codex and the API.
ChatGPT can now create animated GIFs from nothing more than a text prompt: describe the scene and the motion you want, without having to upload a photograph or assemble the frames yourself.
OpenAI introduced the update on November 4, 2026 for ChatGPT, Codex and the API. The image model is available worldwide across all ChatGPT subscription tiers, while GIF creation is rolling out gradually to users.
The feature arrives with Images 2.6, which OpenAI says generates images 50% faster while keeping edits more consistent. Below, two GIFs created from short instructions show a pelican at Lake Kerkini and an excited cartoon owl.
https://openai.com/index/introducing-images-2-6/?utm_source=chatgpt.com&video=1243363300
How to make a GIF using only a text prompt
Simply describe the scene, what should move and whether the motion should repeat as a seamless loop. ChatGPT generates the scene and the sequence of frames. No starting image is required: you can create a GIF from scratch using text. Uploading a still picture or rough sketch is an optional alternative.
OpenAI showcases everything from changing expressions on animated characters to simple motion for mascots, reaction images and storyboards. A short loop can work as social content or a quick movement study before producing a full video.
Example prompt: “Make an animated GIF of a small owl on a branch that gets excited, opens its wings and makes a cheerful hop. Fixed camera, gentle motion and a seamless loop.” You do not have to provide a photograph of the owl to begin.
If the result needs refining, give another instruction, such as “slow down the movement” or “keep the background completely still”. You can also adjust playback speed and save the generated GIF.
Two real GIF examples from simple prompts
We gave ChatGPT two very short instructions without supplying a starting photograph: a pelican on a boat canopy at Lake Kerkini and an excited owl. Each of the two resulting GIFs has nine frames and plays on a loop.
Example 1 — A pelican at Lake Kerkini. The original instruction was simply: “Make a pelican at Lake Kerkini standing on the canopy of a boat.” ChatGPT produced a Dalmatian pelican spreading its wings, with the lake, boat and mountains in the background.
Refined prompt: “Create a short animated GIF featuring a Dalmatian pelican at Lake Kerkini in Greece, standing on the blue fabric canopy of a boat. The pelican spreads its wings and gently turns its head. Keep the boat, lake and mountains in the background stationary, with no camera movement. Make the last frame transition smoothly back to the first for a seamless loop.”
Example 2 — An excited owl. The second instruction was “Make a GIF of an owl looking excited.” ChatGPT produced an illustrated owl that spreads its wings, hops off a branch and returns to its starting pose against a fixed night scene.
Refined prompt: “Create a cute animated GIF featuring an owl perched on a tree branch beneath a bright full moon. The owl becomes excited, widens its eyes, raises its wings and makes a small joyful hop before returning to its starting pose. Keep the branch, moon and forest background stationary. Make the animation loop smoothly without text.”
The useful details are what should move, what should stay still, and how the final frame connects back to the first. These are short loops, not substitutes for a fully produced video scene.
50% faster image creation and more consistent editing
OpenAI says Images 2.6 generates images 50% faster than the previous version. For creators, the advantage is not only shorter waiting times: it means testing more variations of the same idea while better preserving its underlying design.
The company highlights greater consistency for people, products and brand elements across repeated edits. This builds on ChatGPT Images 2.5 and its more precise image-editing workflow, now with greater emphasis on speed and motion.
Flare, Sunburst and new controls through the API
For developers, GPT-Image-2.6 Flare prioritizes speed and, according to OpenAI, can generate an image in as little as six seconds at low quality. Sunburst targets more detailed work and stronger visual consistency across edits, especially at medium or high quality.
The API also gains image markup for pointing directly to the part of a picture that needs changing, SVG output for use in web projects, apps and print, and a complexity parameter that controls how much planning the model applies to a request. Image quality influences visual fidelity, while complexity governs how much effort goes into constructing the scene.
Availability and what users need to know
Images 2.6 is available through ChatGPT on web, desktop and mobile across all subscription tiers, as well as in Codex. GIF creation is being rolled out progressively, so it may not yet appear for every user. Image-use limits depend on the user’s plan.
Another addition is saved reference images: a user can name and save a visual reference, then call it up in a later chat without uploading it again. Flare and Sunburst are also available via the API, which has separate usage-based billing.
More information here.
What we think
The appealing part is that you can type two sentences and get a moving image without finding a photograph, designing a character or opening animation software. For a social post, a short event teaser or a quick visual joke, that is genuinely useful.
Our two examples also show why clear motion instructions matter: the pelican needs to stay on the canopy and the owl has to return to its starting pose for the loop to work. The clearer the distinction between moving and stationary elements, the more useful the result becomes.
A GIF is not a substitute for fully produced video with editing and sound. For a short repeating movement, however, starting directly with a written instruction instead of building frames yourself makes it much easier to test an idea quickly.
Frequently asked questions
Do I need to upload a photo to create a GIF?
No. You can describe the scene and its movement entirely in text. Uploading a still photograph or sketch is an optional starting point.
Is GIF creation available to free ChatGPT users?
Images 2.6 is available across all ChatGPT tiers, but the GIF feature is rolling out progressively. Access and limits may vary by plan and rollout status.
How can I make a GIF loop smoothly?
Ask for a “seamless loop” and describe small changes that bring the final frame naturally back to the starting pose. Specify a fixed background or camera when necessary.
