ChatGPT Images 2.6 can now make animated GIFs

ChatGPT can now make animated GIFs. OpenAI announced ChatGPT Images 2.6 on November 4, saying it adds “GIFs and reference images,” and released two new image models for developers alongside it.

OpenAI’s announcement on X, which includes a video, reads:

Images, upgraded—introducing ChatGPT Images 2.6.

Now with GIFs and reference images.

Plus, two new models are coming to the API with 50% faster image generation.

— OpenAI (@OpenAI) November 4, 2026

OpenAI’s YouTube channel carries the launch video:

What is confirmed for ChatGPT

OpenAI’s post and video say ChatGPT Images 2.6 adds GIF generation and reference images. That is all OpenAI has said publicly about the ChatGPT side that we could find. OpenAI’s ChatGPT image documentation had not been updated for GIFs or version 2.6 when we checked on November 5, and it says only that image availability and usage limits depend on your plan. So we cannot confirm which ChatGPT plans have GIF generation, how long a GIF can be, or what the limits are. Check inside the app before planning around it.

What developers get: Flare and Sunburst

OpenAI’s API changelog lists the release on November 4: GPT-Image-2.6 Sunburst and GPT-Image-2.6 Flare, for image generation and editing “with adjustable complexity, SVG generation, and animated GIF generation.” The tweet says the models are “coming to the API,” but the changelog and model pages list them as released, so we follow the documentation.

OpenAI describes them this way:

• Sunburst: image generation and editing for detailed visual work.
• Flare: image generation and editing for fast iteration.

OpenAI’s video on the API models is titled “GPT-Image-2.6: Faster image generation with Flare and Sunburst.” The “50% faster” claim in the tweet is not tied to a comparison model in the post, so treat it as a headline figure until OpenAI publishes the details.

New controls, according to OpenAI’s image generation guide:

• Animated GIFs. Set output_format to gif on the image generation endpoint and describe the subject, motion and loop in the prompt. You can set frame_count from 2 to 64 frames (default 9) and frame_duration_ms from 20 to 2,000 milliseconds per frame (default 200). GIFs work for generating images, not for edits, and do not support streaming or partial images.
• SVG output. Set output_format to svg for editable vector graphics such as icons and logos. OpenAI says the file contains static vector content, with no scripts, external resources or embedded raster images, and one image per request.
• Complexity. A complexity setting of low, medium or high controls how much the model plans before rendering. It defaults to high, adds no separate fee, and higher settings can raise latency and cause unwanted changes.
• Markup-guided edits. You can mark where a change should happen on an overlay and describe it in the prompt. This works for supported GPT Image 2.5 and 2.6 models, at the same token rates as other edits.

OpenAI’s guide shows this official example: a GIF made by Sunburst with 9 frames at 200 milliseconds each, in which a monkey in a green coat lifts a glowing lantern and lowers it again in a loop of 1.8 seconds.

Price and limits

OpenAI’s model pages list image output at $30 per million tokens for both models, with text input at $5 per million tokens and image input at $8, and cheaper cached rates for repeated input. Costs for SVG are split: you pay the image model’s rate for the reference image and the authoring model’s rate for the tokens that produce the SVG.

OpenAI’s guide also lists limits that still apply: complex prompts can take up to two minutes, text rendering can still fail on precise placement, characters can look different between generations, and layout-sensitive compositions can be hard to control. It also notes that you may need API Organization Verification before using GPT Image models.

What this means for working professionals

• Marketers and social teams: short looping GIFs for posts and email are now one prompt away, so tests are cheaper to run.
• Teachers and presenters: simple animated diagrams are possible, but check every frame for accuracy.
• Designers and small businesses: SVG output gives editable shapes for icons and logos. OpenAI’s own guide shows bakery logos generated this way. Treat them as drafts, not finished brand identities.
• Anyone publishing the results: generated images and GIFs are filtered under OpenAI’s content policy, and you remain responsible for what you post.

We have not tested Images 2.6 ourselves. Everything above comes from OpenAI’s announcement, videos and documentation, and the ChatGPT-side details may change as OpenAI updates its pages.