OpenAI Launches ChatGPT Images 2.5 With Sketch Tools and New API Tiers

OpenAI launched ChatGPT Images 2.5 on Tuesday, an updated model that reduces generation latency by up to 50 percent compared to Images 2.0. [1][2] The system includes a new Sketch feature for drawing reference images and allows users to place comments directly on images for targeted editing. [1]

The upgrade is available across web, mobile, and desktop for all ChatGPT, ChatGPT Work, and Codex users. [1] Developers receive access to two new API models: the faster GPT-Image-2.5 Flare, and GPT-Image-2.5 Sunburst, which is designed for detailed work and has longer generation times. [1]

The model is designed to better preserve the likeness of people and pets in reference photos. [2] Early testing demonstrated the model's ability to maintain a specific subject's appearance across different visual styles, such as turning a cat reference photo into a Renaissance portrait. [2]

Users can also utilize built-in templates for common formats like product photos and flyers. [1]

The introduction of ChatGPT Images 2.5 focuses on workflow integration and precision control through visual commenting and sketching. By offering two distinct API models—Flare for speed and Sunburst for detail—OpenAI is segmenting its developer offerings to support both rapid prototyping and intensive design work. What the sources don't address: How the pricing for the new Flare and Sunburst API models compares to previous GPT-Image tiers.