OpenAI ships ChatGPT Images 2.5 with Sketch and faster API models

OpenAI published ChatGPT Images 2.5, saying it offers sharper detail, more precise multi-turn edits, and up to 50% less wait than Images 2.0.

ChatGPT can now draw from a sketch you make in the chat, and developers get a faster image model. The 50% wait cut and the 3-billion-a-week volume are OpenAI’s numbers.

OpenAI published “Introducing ChatGPT Images 2.5.” That product page is the filing event. OpenAI claims sharper details, more precise multi-turn editing — follow-up edits in the same chat — and “50% lower latency,” meaning about half the wait, versus Images 2.0 on the faster API line. Those quality and wait figures are OpenAI’s.

OpenAI says it is available to all ChatGPT, ChatGPT Work, and Codex users on desktop, mobile, and web. The Verge’s same-day recap repeats that rollout.

New ChatGPT tools on the product page: Sketch — draw directly in ChatGPT as a reference for the final image (The Verge: type @Sketch to open the drawing window); templates for flyers and product photos; comments placed on images for focused edits; and shareable prompts so others can retry an idea with their own photos and details.

For developers, the API — the programming hook — has two Images 2.5 options. Flare is the default and faster path: OpenAI says higher-quality images than the last version at 50% lower wait. Sunburst is the higher-precision option and takes longer. Model IDs and the dated Flare snapshot are in Sources.

OpenAI writes: “Every week, people create more than 3 billion images across ChatGPT Images and the GPT-Image models in the API.” That volume figure — three billion images a week — is the company’s.

Safety on OpenAI’s Images 2.5 system card, published the same day: C2PA metadata (C2PA is a standard for labeling how media was made), plus invisible watermarking through Google DeepMind’s SynthID on ChatGPT, Codex, and the API. Those are OpenAI’s stated controls, not a third-party audit.