OpenAI Launches ChatGPT Images 2.5 With Faster, More Precise Editing
OpenAI introduced ChatGPT Images 2.5 on September 8, 2026, its newest image generation model, promising better quality, more reliable multi-turn editing, and generation speeds up to 50% faster than Images 2.0.
Details
- Quality gains: more natural lighting and richer textures, plus better preservation of subjects pulled from reference photos
- Faster generation: latency cut by as much as 50% compared with the previous Images 2.0 model
- More reliable editing: follows instructions more consistently across multiple rounds of edits while keeping the image coherent, handles transparent backgrounds, and holds requested visual styles
- New creation tools: a Sketch feature for drawing directly in ChatGPT as a visual reference, ready-made Templates for common outputs like flyers and product photos, Image Comments for pointing at specific spots to request edits, and Shareable Prompts so others can remix a given image’s prompt
- New developer models: GPT-Image-2.5 Flare, tuned for speed and general use, and GPT-Image-2.5 Sunburst, a premium option for high-precision creative work
- Availability: rolling out immediately to ChatGPT, ChatGPT Work, and Codex users across desktop, mobile, and web
What happened next
The update lands about a week after OpenAI retired the standalone DALL-E GPT from ChatGPT on August 30, continuing the consolidation of image generation and editing into ChatGPT Images as the single in-product destination. Developers gain the two new GPT-Image-2.5 API models the same day, giving them a speed-optimized and a precision-optimized option to build on going forward.