OpenAI announced GIF creation in ChatGPT on November 4, 2026, as part of Images 2.6. The new output gives everyday users, creators and marketing teams a way to make short, looping animations alongside still pictures.
OpenAI’s examples include animated reactions, moving characters and product loops. The company also says the update generates images faster and keeps character and product details more consistent during edits.
Access differs between ChatGPT and Codex
OpenAI’s announcement describes global availability across ChatGPT and Codex subscription tiers. But Codex’s current pricing page explicitly excludes image generation on Free, so the launch should not be read as a promise of free GIF creation in Codex.
Codex image generation also consumes included usage roughly three to five times faster than comparable turns without images, depending on image size and quality, according to OpenAI. A GIF-specific ChatGPT daily allowance and maximum animation length are not established in the launch announcement or ChatGPT help guidance.
Check more than the movement
Improved editing is not a guarantee that everything else in an image stays unchanged. ChatGPT’s general image-editing guidance says edits can affect more than the selected region. The API guide also lists continuing difficulties with precise text, brand consistency and exact layouts.
For a product loop, the practical implication is to review the whole animation—including lettering and product features—not only the requested movement.
What changes for teams automating GIF creation
Teams building GIF creation into apps can use the API, OpenAI’s programmatic interface. Its guide allows 2–64 frames, subject to size limits and a shared pixel budget, but the image-editing endpoint cannot output GIFs. These are API restrictions, not published limits for ChatGPT’s interface.
For scale, the documented defaults are nine frames at 200 milliseconds each. Calculation: assuming each frame plays once per cycle, 9 × 200 ÷ 1,000 gives a 1.8-second loop. That is playback duration, not generation time.
Both API models, Flare and Sunburst, list standard US-dollar prices of $8 per million image-input tokens and $30 per million image-output tokens, plus $5 per million text-input tokens. Tokens are the units used to meter the request and result; these rates are not a flat charge per GIF.
API usage is billed separately from ChatGPT usage, although eligible Pro and Business subscribers can activate credits that cover Images API calls. Eligibility, account-linking and expiry rules apply. The credits do not raise ChatGPT or signed-in Codex limits, and do not cover Codex usage through API keys.
For a marketing budget, compare the cost of a usable final loop—not one generation attempt. Include rejected attempts and review time, as RohitAI’s guide to the cost of an approved AI image explains.