Updates·November 4, 2026, 20:44

ChatGPT Images 2.6 makes GIFs and remembers your pictures

AI-generated from the sources listed, with automated source checks.

OpenAI has launched Images 2.6 in ChatGPT, letting users create animated GIFs and save reference images, while two new image models are coming to developers.

AI-generated image

What's new

OpenAI has launched ChatGPT Images 2.6. The biggest addition is the ability to create an animated GIF from a text message or an image you upload yourself. Afterward, you can adjust the speed, edit individual frames in the animation and share the GIF so it plays in a loop.

OpenAI demonstrated the prompt "Can you make a GIF of a cat dancing?" The model creates several small variations of the character and stitches them together, like a flipbook cartoon. You can also save reference images of yourself, your pets and characters and reuse them without uploading them every time.

For developers, two new API models are on the way (an API is the access point other programs use to talk to the service): GPT-Image-2.6 Flare and GPT-Image-2.6 Sunburst.

What's smart about it

The smart part is that the character keeps looking like itself from frame to frame. According to The Decoder, the model's consistency looks strong enough to keep the character recognizable, while the publication believes older models would have changed its appearance too much along the way.

The saved reference images make it easy to create more content with the same dog, the same cartoon character or yourself.

According to OpenAI, Flare can generate an image in as little as six seconds, and Sunburst delivers higher quality and greater consistency across edits. Both models support image markup and export as SVG, an image format that can be scaled without becoming blurry. According to OrcaRouter, the six-second figure applies to low quality.

Cheaper or better?

OpenAI says image generation is about 50 percent faster. That is the company's own measurement, and there are still no independent tests of 2.6.

For developers, the third-party blog OrcaRouter reports the following API prices: $5 per million text input tokens, $8 per million image input tokens and $30 per million image output tokens. Tokens are the chunks of text that pricing is based on. An image costs about half a cent at low quality in 1K size and roughly $0.05-0.10 at high quality. The prices do not come from OpenAI itself.

OrcaRouter mentions Google Gemini and Imagen, xAI's Grok Imagine Image and the predecessor GPT-Image-2.5 as points of comparison, but no direct price or quality comparison is provided. No separate price has been given for regular users, but according to the sources, Images 2.6 is rolling out to all subscription tiers in ChatGPT and Codex on web, desktop and mobile. The sources do not say whether this applies in Denmark and the EU.

What it's good for

For fun and personal projects: a dancing cat, a GIF of a cartoon character for social media or a greeting starring your own pet. With saved reference images, you don't have to start over every time.

For businesses and developers, GIF is now an output format in the API alongside PNG, JPEG and WebP, and a complexity parameter (low, medium, high) can control the level of effort. Flare suits cases where speed matters, while Sunburst suits images that need to be edited multiple times and still look the same. SVG export is handy for logos and illustrations.

OpenAI says users are now generating well over four billion images a week via ChatGPT and the API, a billion more than in September. The figure is OpenAI's own.

Sources

More on this topic

Get the week's AI news in your inbox

Choose your level, topics and length. One email a week, unsubscribe at any time.

Subscribe to Promptly Newsletter