OpenAI Adds Sketch to ChatGPT Images 2.5, Ships Two API Models

OpenAI on September 8 released ChatGPT Images 2.5, positioned as its new state-of-the-art image model, alongside two API variants and a hand-drawn sketch input feature. Per the company's launch post, image generation latency drops "by up to 50% compared with Images 2.0," and the update rolls out to ChatGPT, ChatGPT Work, and Codex users across all tiers on desktop, mobile, and web.

The Sketch feature, invoked by typing "@Sketch" in a chat, "lets users draw directly inside ChatGPT as a visual reference for the final image." New templates provide starting points for popular formats, with named options including "Poster" and "Merch," and users can place comments on generated images for focused editing. 9to5Mac summarized the release as delivering "sharper details, faster generation, more precise editing, and better tools for creating and sharing."

For developers, Unite.AI reports two API models. GPT-Image-2.5 Flare is the default, described as "OpenAI's fastest model for high-quality, everyday image generation," and delivered "high-quality images at two to four times the speed of GPT-Image-2" in OpenAI's evaluations. GPT-Image-2.5 Sunburst is built for "workflows where editing precision matters most," aimed at production-ready campaign creative and polished product imagery with longer generation times. Both are priced at $8.00 per 1M image input tokens and $30.00 per 1M image output tokens.

OpenAI cites safety metrics of 1.09% unsafe generations for Sunburst and 1.41% for Flare versus a 1.64% baseline. Users, the release notes, create more than 3 billion images each week on the platform. It lands as our fourth OpenAI item of the day, alongside earlier coverage of Altman on coding agents at YC.