OpenAI has begun rolling out ChatGPT Images 2.5, its latest image-generation model, combining improvements in both speed and quality to make AI image creation more precise and seamless.
اضافة اعلان
Users already generate more than 3 billion images per week through ChatGPT and the API, and the new update is designed to speed up that process while noticeably improving its accuracy.
What Sets ChatGPT Images 2.5 Apart?
The biggest improvements include sharper detail, more natural lighting, and textures that look more realistic rather than artificial.
The model is also better at staying faithful to reference images. When using a personal photo and transforming it into a new scene or visual style, facial features and key characteristics are meant to stay more consistent and recognizable, rather than shifting significantly during generation.
OpenAI said the new model is now faster, more precise, and smarter, with better image-creation tools, pointing to a set of improvements including faster image generation, sharper facial accuracy for more natural results, consistent detail preservation across multiple edits, and new tools for making edits through comments.
Image editing has also become smarter. When a user requests a single change, like swapping out the background or adjusting a product within the image, the model is meant to keep the rest of the image's elements unchanged.
This consistency also carries across multiple rounds of edits, so previous changes don't randomly disappear when new instructions are entered.
On the speed front, image generation is now up to 50% faster compared to ChatGPT Images 2.0.
Sketch Your Idea, and AI Finishes It
OpenAI has added a new feature called Sketch, allowing users to draw their idea directly within ChatGPT, whether it's a room layout or a clothing concept, with the AI then turning the rough sketch into a polished final image.
The feature comes alongside new templates for posters, ad posts, and product images, as well as the ability to leave comments directly on images for more precise edits.
These tools give users greater control over the final result, without needing prior design experience.
Users can now also share the original text prompt used to create a specific image. So if someone creates an image another user likes, that user can reuse the same prompt to recreate the idea with their own photos and details.
For developers, OpenAI is also releasing the GPT-Image 2.5 Flare and Sunburst models via the API, bringing the same speed and quality improvements to applications and products built on the company's technology.
ChatGPT Images 2.5 is now available to all ChatGPT, ChatGPT Work, and Codex users across desktop, mobile, and web.
