OpenAI announced on December 17 that its main image generation model, GPT Image 1.5, would be used in a new version of ChatGPT Images. This is a big improvement in how AI users make and change images.
The update is now available to all ChatGPT users and API buyers around the world. OpenAI promises that the update will make creating images faster, change them more accurately, and keep more of the original image’s features.
Is there anything new in ChatGPT Images?
The newest version of ChatGPT Images lets users make images from scratch or make much more accurate changes to photos they already have. OpenAI says the model can make pictures up to four times faster while keeping important parts of the pictures the same across changes. These include lighting, framing, and how people look.
With the new Images tool in ChatGPT, there is now a dedicated creative space. Preset styles, filters, and hints make it easy to explore images.
More accurate editing of pictures
Following directions better is one of the biggest changes. When a user edits a shared picture, the model now only changes the parts that the user asks it to and leaves the rest alone. This lets you:
- Photo changes that look more real
- Realistic hair and clothes try-ons
- Cleaner filters for style and changes to ideas
The model lets you do complicated editing tasks like adding, deleting, mixing, joining, and transposing parts without changing the quality of the original picture.
More innovation and better text rendering
OpenAI says the model is very good at creative changes like adding text, changing patterns, and rethinking images while keeping the main idea of the original image. Text modeling has also gotten better, which means the AI can handle both thick and thin text more correctly.
Better handling of multiple small faces and more natural-looking outputs are two more quality improvements that make pictures more useful in the real world.
It costs less and makes more with GPT Image 1.5 in the API.
With the GPT Image 1.5 API, you can get the same changes. According to OpenAI, this version is much better at keeping images and brands consistent than the last one. Because of this, it’s great for:
- Photo changes that look more real
- Realistic hair and clothes try-ons
- Cleaner filters for style and changes to ideas
OpenAI has also cut the costs of both inputting and outputting pictures by 20% compared to GPT Image 1. This means that coders and companies can make more images with the same budget.
OpenAI says that GPT Image 1.5 is already being used by businesses and companies in artistic tools, e-commerce, marketing software, and other fields. Developers can use the OpenAI Playground to try out the model or the official help guide to learn about best practices.
How to get it and what’s next
Starting today, the new ChatGPT Images model is available everywhere through ChatGPT and the API. You don’t have to choose the model by hand. You will still be able to get the older version of ChatGPT Images as a special GPT.
OpenAI says that this release is a big step forward in AI picture creation. In the future, updates will likely add more language support, finer-grained editing and better visuals.
