On November 4, 2026, OpenAI introduced Images 2.6 in ChatGPT, Codex and the API, available globally across all ChatGPT subscription tiers. OpenAI says it generates images 50% faster than before and can now produce animated GIFs. In the API it comes as two models: GPT-Image-2.6 Flare, which can produce an image in as little as six seconds at low quality, and GPT-Image-2.6 Sunburst, aimed at higher-quality output and greater consistency across edits.
The API additions include image markup (pointing at exactly where an edit should go, already in ChatGPT), SVG export, and a new complexity parameter that sets how much planning the model does, separate from the existing quality setting. The API changelog adds that markup-guided editing also works on supported GPT Image 2.5 models. Sunburst is priced at $5 per million text input tokens, $8 per million image input tokens and $30 per million image output tokens, and works only on the image generation and edit endpoints.
OpenAI also says people now create 4 billion images per week across ChatGPT, Codex and the API, more than half a billion a day. Launch partners quoted include OpenArt, Figma Weave, Picsart and Town.
Why it matters: image models are moving from single pictures toward production assets - vector files, short animations and controlled multi-round edits - which is what design and marketing tools actually need to embed.
What it does not show: the speed figure and the 4 billion weekly images are OpenAI’s own numbers. The announcement gives no quality benchmark against other image models, and safety details are deferred to a separate system card.