OpenAI launches ChatGPT Images 2.5 with faster generation and better editing

OpenAI has announced an image generation model called ChatGPT Images 2.5 which follows the existing GPT-image-2. The company says that this new model is mainly designed to create images, produce sharper pictures and offer tighter editing control. Developers will notice a change: GPT-Image-2.5 Flare, the standard model used by most API apps, produces higher‑quality images than GPT-Image-2 and can cut waiting time by as much as fifty percent. In short, you can expect images to appear quicker while staying just as good.

In a blog post OpenAI says that people make, then 3 billion images every week using ChatGPT Images and the GPT-Image models that work through an API. Now with these models the AI company wants to make images that look more natural.

Natural-Looking Images

It all comes down to better lighting, texture, and levels of detail—areas for which these new models were designed. They are also capable of preserving the distinctive features of people and objects when used as reference images.

For example, if you upload a photo of a person and ask ChatGPT to place the person in a different setting or visual style, Images 2.5 does a better job of keeping the person's recognisability.

More Precise Editing

One of the useful changes according to the company is the ability to make small adjustments without changing the rest of the picture.

Do you want to change the background while keeping the person intact? Or replace a product while maintaining the same composition? Image 2.5 is designed to follow those instructions with greater fidelity. The model also improves how it handles multiple edits within a single conversation. This means that if you request successive changes, previous edits are more likely to remain intact, preventing the image from gradually morphing into something you never asked for.

Now you can sketch out your idea

Explaining something in words isn't easy for everyone; sometimes it can be tricky. But there are other ways to explain things—one of which is simply by drawing. OpenAI has introduced this capability through a new "Sketch" feature in ChatGPT.

Users can draw a quick sketch directly in ChatGPT and use it as a visual guide for the final image. It could be the basic layout of a room, an outfit design, or even a rough doodle that, in a way, only you understand.

You can then add a text description to specify the style or other details. To try it out, simply type "@Sketch" in ChatGPT.

Two new models for developers

OpenAI is also adding two image models to its API

GPT-Image-2.5 Flare is the default choice for most applications; it offers new quality and editing improvements, as well as up to 50% lower latency compared to GPT-Image-2.

Then there is GPT-Image-2.5 Sunburst, geared toward creative work that requires greater detail. Although it takes longer to generate images, it is designed to provide more precise control over complex edits.

For regular ChatGPT users, Images 2.5 is rolling out across desktop, mobile, and web versions. In essence, this update streamlines image creation and makes editing more predictable, giving users a new way to explain an idea when words simply aren't enough.

ABOUT THE AUTHOR

Kahekashan

[email protected]

Kahekashan is a passionate technophile with a keen eye for cutting-edge gadgets, emerging technologies, and everything in the digital realm. Raised in a Defence family with strong values and a background in literature, she has consistently pursued excellence in every endeavour. Her last full-time assignment involved content writing with the Indian School of Business.

Read More