OpenAI promises faster, more precise photo editing with ChatGPT Images 2.5

OpenAI said the ChatGPT Images 2.5 improves image quality, reference-photo fidelity, and editing consistency, while cutting generation latency by up to 50 per cent

OpenAI has launched ChatGPT Images 2.5, its latest image-generation model, with improvements in image quality, editing accuracy and generation speed. According to the company, the new model can produce sharper details, more natural lighting and richer textures, while better preserving subjects from reference photos. Image-generation latency has also been reduced by up to 50 per cent compared with Images 2.0.

ChatGPT Images 2.5: What's new

Better image quality

 

OpenAI said the ChatGPT Images 2.5 is designed to produce more natural-looking images, with improvements to lighting, textures and fine details. The company said the model is also better at retaining distinctive features when users provide reference photographs. This means users can use an existing photo as a reference and transform the subject into different settings, visual styles or compositions while keeping it recognisable. Try ChatGPT 80s photo prompt

 

More precise editing

The new model also focuses on making individual changes without altering the rest of an image, said OpenAI. Users can ask ChatGPT to modify a specific element, such as a product, background or piece of text, while preserving the subject, composition and other details. OpenAI said this is particularly useful for more complex images, where changing one element can otherwise affect the entire composition.

Consistent edits across conversations

Images 2.5 is also designed to maintain consistency through multiple rounds of editing. According to OpenAI, earlier changes are more likely to remain intact when users continue refining an image in the same conversation.

 

This allows users to make several targeted edits without having to recreate the image from scratch or seeing its quality degrade with each change. 

Generation is faster

OpenAI has also focused on reducing the time required to generate images. The company said Images 2.5 cuts generation latency by up to 50 per cent compared with Images 2.0.

 

The faster generation is aimed at making the process of creating and refining images more interactive, particularly when users are experimenting with multiple versions of the same idea.

New tools and API models in ChatGPT

 

Alongside the model, OpenAI is introducing new features to give users more control over image creation.

 

Sketch allows users to draw directly in ChatGPT and use the drawing as a reference for the final image. The company is also adding Templates, which provide starting points for formats such as flyers and product photographs.

 

Users can now place comments directly on images to indicate areas that need editing. They can also share the prompts used to create an image, allowing others to try the same idea with their own photos and details.

 

OpenAI is also bringing the Images 2.5 improvements to developers through two new application programming interface (API) models.

 

GPT-Image-2.5 Flare offers the same improvements in image quality, editing and speed. GPT-Image-2.5 Sunburst is designed for more detailed creative work, offering greater precision but taking longer to generate images.

 

The improved reference-photo handling is also aimed at production workflows, where developers may need to create variations while keeping the original subject and visual treatment consistent.

 OpenAI introduced ChatGPT Images 2.0 as an earlier generation of its image model, focused on improving image creation and editing within ChatGPT. Images 2.5 builds on that foundation with a stronger focus on reference-photo fidelity, precise editing, consistency across multiple edits and faster generation. 

 

Availability

ChatGPT Images 2.5 is available to ChatGPT, ChatGPT Work and Codex users across desktop, mobile and the web. Developers can access the new capabilities through the Images API using the new Flare and Sunburst models.