OpenAI rolls out ChatGPT Images 2.6 with animated GIFs and two new API image models

OpenAI began rolling out ChatGPT Images 2.6 globally across ChatGPT and Codex on web, desktop and mobile across all user tiers, alongside two new developer API models: gpt-image-2.6-flare and gpt-image-2.6-sunburst. According to OpenAI, users create roughly 4 billion images per week and more than 500 million per day across its products.

In ChatGPT, Images 2.6 renders outputs 50% faster than its predecessor while sharpening lighting, skin tones, materials and dense multi-column text layouts across menus, infographics, calendars and slides. Users can save reusable reference images for specific people, pets or characters so faces and outfits stay consistent across sessions without re-uploading files. The model also generates looping animated GIFs directly in chat, converting hand-drawn sketches into short motion loops or assembling six-frame 3-by-2 sprite sheets into smooth animations.

API models and developer controls

On the OpenAI API, gpt-image-2.6-flare prioritises low-latency iteration, returning low-quality drafts in as little as 6 seconds while improving poster and product typography. gpt-image-2.6-sunburst targets higher photorealism, denser spatial composition and stricter subject preservation across multi-turn edits at medium and high quality settings. Both endpoints add markup-guided region editing, native vector SVG output for icons and diagrams, and a three-level complexity planning parameter (low, medium and high) that scales upfront layout reasoning before pixel synthesis.

Key numbers

• $8 per 1M image input tokens ($2 cached) and $30 per 1M image output tokens for both Flare and Sunburst
• $5 per 1M text input tokens ($1.25 cached) on both API image endpoints
• Up to 6-second generation latency at low quality on gpt-image-2.6-flare
• 50% faster end-to-end image rendering inside ChatGPT compared with the prior model