Google threw down the AI image gauntlet last month. Now, OpenAI has answered.
The ChatGPT creator announced the rollout of a new flagship image generator on Tuesday, which the company says is capable of making faster, more precise edits to an AI image while maintaining details.
But the more eye-catching change for many user will likely be the improvements to image quality and the AI model's ability to follow specific instructions.
OpneAI said the new model offers "clear improvements across a range of cases." In one example, the company showed off the differences between the old and new image model when prompted to generate a photorealistic scene in 1970s Chelsea, London.
In another example touting use cases for businesses using the company's image API, OpenAI compared the outputs showing a mechanic working on a car.
While a model's ability to generate photorealistic images has become a popular point of comparison in the AI race, OpenAI's latest model can also generate animated images, graphics, and other styles of artwork.
Perhaps to boost awareness of that, the company is rolling out a new Images feature within the ChatGPT app. While AI image generation was previously available within ChatGPT, OpenAI says the new dedicated Images feature is designed to "spark inspiration and make creative exploration effortless."
The news comes just over three weeks after Google released its Nano Banana Pro AI image model alongside its flagship Gemini 3 LLM — both of which have received widespread praise and reignited the debate around whether Google had begun to overtake OpenAI in the AI race.
Google's new AI image generator was lauded for its hyper-realistic AI images, which some people used over Thanksgiving to make it appear as if they had famous guests at the holiday dinner table.
Yuchen Jin, the cofounder and CTO of the startup Hyperbolic Labs, called OpenAI's new model "Nano Banana Pro level in my tests."
