Technology
OpenAI Rolls Out ChatGPT Images 2.5 With Faster, More Precise Editing Tools
OpenAI has launched ChatGPT Images 2.5, a new image-generation model the company says offers sharper detail, more accurate editing and faster performance than its predecessor. The update is accompanied by two new models for developers using OpenAI's API.
The rollout coincides with a viral trend in which users transform selfies into 1980s-style portraits using a prompt called "Neon 80s Portrait," which reads "@Create image Show me what I would look like if I was in the '80s." This trend has been driven by a new prompt-sharing feature included in the update, which allows users to share the exact prompt behind a generated image so others can apply it to their own photos.
Alongside prompt sharing, OpenAI introduced a tool called Sketch, accessed by typing "@Sketch," which lets users draw directly within ChatGPT and use that drawing as a reference for a finished image. The update also adds templates for formats like posters and merchandise, as well as the ability to leave comments directly on an image to request specific edits.
OpenAI said users generate more than 3 billion images each week across ChatGPT Images and the GPT-Image models available through its API. According to the company, Images 2.5 reduces image generation latency by up to 50 percent compared with Images 2.0.
The company said the new model is better at preserving the likeness of people from reference photos when placing them into new settings or styles, and produces more natural lighting and textures. It also said Images 2.5 is more effective at applying requested edits without unintentionally altering other parts of an image, and that this accuracy holds up across multiple rounds of editing within the same conversation, rather than declining as more edits are made.
ChatGPT Images 2.5 is being made available to all ChatGPT, ChatGPT Work and Codex users across desktop, mobile and web platforms.
For developers, OpenAI is releasing two new models through its Images API. GPT-Image-2.5 Flare includes the same improvements in quality, editing and speed, and is positioned as the default option for most use cases. OpenAI said it produces higher-quality images than GPT-Image-2 while cutting latency by 50 percent, making it suitable for applications such as social media content, product imagery, visual search and high-volume image generation.
The second model, GPT-Image-2.5 Sunburst, is designed for premium workflows that require more precise control over edits, such as campaign creative and polished product imagery. It has longer generation times than Flare.
Regarding safety, OpenAI said the new model retains existing safeguards, including checks on prompts and images designed to prevent harmful content. The company said it continues to apply C2PA metadata and invisible watermarking to all images generated using its tools, and referred to a system card for additional details on its safety evaluations.