OpenAI announces GPT-4o’s powerful new image generation tool: Here’s what it can do

OpenAI

OpenAI just dropped its most advanced image generation tech yet, and it’s now built right into GPT-4o. That means you can create super-detailed and realistic images just by typing out what you want—no fancy tools or design skills needed.

For a while now, OpenAI has been experimenting with AI-generated images, but this update takes things to a whole new level. Whether you need a simple sketch, a complex infographic, or even a comic strip, GPT-4o can handle it. It can even generate text inside images—something older models really struggled with.

One of the coolest parts? You can tweak your images through conversation. Start with a rough idea, then refine it step by step until it looks exactly how you imagined. This is especially handy for designers, educators, and creatives who need precise visuals.
Another big improvement is how it handles text inside images. If you’ve ever seen AI-generated images with weirdly placed or unreadable words, you’ll know this has been a problem. GPT-4o fixes that, making it much easier to create posters, graphics, or illustrated stories with actual readable text.

It’s also gotten way better at following instructions. If you want an image with multiple objects, specific colors, or fine details, GPT-4o can actually deliver—something other AI models often mess up.

And it doesn’t stop at just generating images. GPT-4o can also analyze pictures you upload. So if you want to maintain a consistent style across a project, you can provide reference images, and the AI will adjust accordingly. It even understands historical and cultural details, making sure visuals are accurate and relevant.

Of course, no system is perfect. OpenAI is still working on safety features to prevent misuse, and every image comes with metadata so people know it was AI-generated.

Starting today, GPT-4o’s image generation is rolling out to Plus, Pro, Team, and Free users in ChatGPT. Enterprise and educational users will get access soon, and developers will be able to use the API in the coming weeks.

If you’re a fan of OpenAI’s previous image generator, DALL·E, don’t worry—it’s still around. But with these new upgrades, creating images with GPT-4o is easier, smarter, and more flexible than ever.