OpenAI made its image generation offerings more precise and consistent in its latest update to ChatGPT Images, as more enterprises and brands use AI image generation to help with design visualization.
The updates will roll out to all ChatGPT users and the API as GPT Image 1.5. The company said it’s powered by GPT 5.2, which many early users found to be a powerful update for business use cases.
“Many people’s first experience with ChatGPT involves turning a text prompt into a picture,” said Fidji Simo, OpenAI CEO of Applications, in a Substack post. “It’s a magical way to see what this technology can do, but the chat interface wasn’t originally designed for this. Creating and editing images is a different kind of task and deserves a space built for visuals.”
Business-friendly updates in precise editing and instruction following
One of the biggest updates to ChatGPT Images is more targeted editing, even when the image is generated on the chat platform rather than through the API. Image generation models such as ChatGPT Images, Google’s Nano Banana, and Stable Diffusion tout prompt-based tweaks to AI-made pictures, where the user can pinpoint specific parts of the photo to change. But those features can sometimes be hit-and-miss.
With the update, OpenAI said the model better adheres to what the user wants “while keeping elements like lighting, composition, and people’s appearances consistent across inputs, outputs and subsequent edits.”
Users can instruct the model to do most types of image editing, such as adding or subtracting an element, combining, blending, and transposing.
OpenAI said that this model “follows instructions more reliably” than previous versions. It’s also able to render text better and generate actual, readable letters, even when these are denser or smaller. OpenAI updated the model to create better, smaller faces in photos featuring a large group of people.
“These transformations work for both simple and more intricate concepts, and are easy to try using preset styles and ideas in the new ChatGPT Images feature — no written prompt required,” according to OpenAI.
Battle of the image generators
OpenAI’s image model update comes after Google’s much-lauded Nano Banana Pro image model, which drew praise from the developer community.
The company must compete with other ever-growing, continually improving image-generation models that aim to attract more enterprise users. And it isn’t just Google that OpenAI has to contend with. In August, Alibaba announced that Qwen-Image can render readable text in both Chinese and English. Black Forest Labs released Flux.2, which also offers a robust, open-source image model.
