On 4 November 2026, OpenAI released Images 2.6, its new image generator, in ChatGPT, Codex and the API. OpenAI says it is 50% faster, can make animated GIFs and, through the API, export SVG files. It matters to anyone who uses AI to illustrate a website, an online store or social posts.
What we know
- Availability: according to OpenAI, Images 2.6 has been available worldwide since 4 November in ChatGPT and Codex, on every subscription tier, on desktop, mobile and web.
- Two API models: GPT-Image-2.6 Flare and GPT-Image-2.6 Sunburst. OpenAI says Flare can produce an image in as little as six seconds at low quality and handles layouts for posters and product images better; Sunburst targets higher quality and better consistency from one edit to the next.
- Speed: the announcement claims 50% faster generation "than before", without naming the baseline version.
- Animated GIFs: available in ChatGPT and the API, for example to animate a brand mascot or turn a product shot into a loop.
- API additions: image markup (pointing at the exact area to change), already in ChatGPT, comes to the API, along with SVG export and a new "complexity" parameter that controls how much planning the model does. OpenAI recommends low complexity for quick edits.
- Provenance: the system card published on 4 November says images still carry C2PA metadata and now also get an invisible watermark, Google DeepMind's SynthID, across ChatGPT, Codex and the API.
- Volume: OpenAI says people create 4 billion images a week across ChatGPT, Codex and the API.
- Pricing: the announcement points to the API pricing page and gives no figures in the text.
What changes and what doesn't
- What changes: animated and vector assets no longer need a designer or a separate tool, so more of them will end up on small-business websites.
- What changes: checking whether an image is AI-made no longer depends only on metadata, which editing or compression can strip; there is now a mark inside the image itself.
- What doesn't change: what search engines need from an image: fast loading, alt text that describes it, and useful content around it.
- What doesn't change: GIF is still a heavy format, whoever or whatever made the file.
How to tell if it affects you
Putting a GIF on a page: web.dev, Google's performance guide, shows a 3.7 MB example GIF shrinking to 551 KB as MP4 and 341 KB as WebM. Lighthouse and PageSpeed Insights flag it as "Use video formats for animated content". Convert it to a short, muted, looping video before uploading, or keep the GIF for social media. Then run the page through PageSpeed Insights and make sure the warning is gone.
Using an exported SVG: an SVG is text and can contain code, not just shapes, which is why WordPress blocks SVG uploads by default. Open it in a text editor or run it through an SVG sanitizer first, and upload it only if your CMS handles SVG safely. The upside: a logo or icon in SVG stays sharp on any screen and is usually small.
Labelling images as AI-made: many compression tools strip metadata, C2PA included. If you want to keep that label, check the final file, not the original. OpenAI offers a public tool to check these signals.
In every case: write the alt text from the final image, not from the prompt, and give the file a name that describes what it shows.
Related: Canva Craft: Canva's First AI Image Model
Sources
- OpenAI, "Introducing Images 2.6", 4 November 2026
- OpenAI, ChatGPT Images 2.6 system card, 4 November 2026
- OpenAI, API pricing (image generation)
- web.dev, "Replace animated GIFs with video for faster page loads"
Updates: if OpenAI changes Images 2.6 pricing or availability, it will be added here with a link.