ChatGPT Images 2.6: Create, edit, and animate images with AI
OpenAI released Images 2.6 on November 4, 2026, across ChatGPT, Codex and its API. The company says it generates images 50% faster than before, and the update is rolling out to every ChatGPT and Codex tier, including Free, on web, desktop and mobile. The headline addition is native GIF creation: a prompt or uploaded still image can become a looping animation, with adjustable speed and editable individual frames.
Reference images are the other consumer change: users can store pictures of themselves, pets or characters and reuse them without uploading again. Image markup lets users point to the exact spots they want changed. Export to SVG produces vector files for websites, apps and print. A new complexity setting controls how much planning the model does; OpenAI suggests low for most quick edits and higher settings for scenes with many elements. Heise reported a gradual GIF rollout, and OpenAI has not disclosed GIF length or frame counts.
For developers, two new models arrive in the API: GPT-Image-2.6 Flare, which OpenAI says can return an image in as little as six seconds at low quality, and Sunburst, aimed at higher quality and steadier results across edits. Per-token prices stayed at Images 2.5 levels, but SVG output costs more because the model renders a raster image first and then converts it. OpenAI says users generate about 4 billion images a week across ChatGPT, Codex and the API, up from 3 billion in September.
Why it matters
For founders building on image generation, the release moves animated loops and vector export into the same model that handles still images. Startups that sold standalone animation or vector tools now compete with a feature that ships to every ChatGPT tier, free included. Reference libraries and markup-guided edits reduce re-prompting, which lowers the effort needed to keep a character or brand asset consistent. Operators should also expect heavier usage, since OpenAI's weekly image count rose from 3 billion in September to 4 billion.
For investors, the pricing details matter. Per-token API rates held steady, but SVG output carries extra cost because it takes a two-step render, so margins on vector deliverables depend on that pipeline. The two model tiers let developers route cheap, fast jobs to Flare and consistency-sensitive work to Sunburst. OpenAI also concedes continued weakness with text, complex layouts and character consistency, so products that rely on those features still need their own quality checks before they promise customers results.
Key points
- 1OpenAI released Images 2.6 on November 4, 2026, across ChatGPT, Codex and the API.
- 2Users can turn a prompt or uploaded still image into a looping GIF and edit individual frames.
- 3GPT-Image-2.6 Flare can return an image in as little as six seconds at low quality, per OpenAI.
- 4SVG export produces vector files for websites, apps and print, with an extra cost in the API.
