Everyone’s favorite chatbot is getting an upgrade.
OpenAI on Tuesday updated ChatGPT to make it easier for users to create and edit images.
It’s also faster—four times faster,
the company says—than OpenAI’s previous image generation model.
You won’t find the words “Google” or “Gemini” or even “Nano Banana” in the blog post announcing the upgrade, but the rival’s fingerprints are all over OpenAI’s announcement.
Still, the update is welcome.
There’s a new “Create image” section in ChatGPT expressly for the medium, rather than forcing you to chat your way to a desired result. Users can make multiple edits to uploaded images, e.g. “turn this into the style of Matisse” or “add a soft light” or even “caption this photo.”
In a test with a photograph of Joseph Gordon-Levitt at our recent Fortune Brainstorm AI gathering, it managed to make all of these edits (even if its take on Matisse was not terribly convincing to a museum curator)—but refused to identify the actor-activist by name, even though it correctly identified him in an earlier exchange.
“The model shows clear improvements across a range of cases, though results remain imperfect,” its maker writes. “While this release represents meaningful progress, there is still significant room for improvement in future iterations.”
—AN 
