Microsoft is coming up with the king of the AI image generator game, and at least one metric, it didn’t miss. On Tuesday the company released two new image-in-picture models, called MAI-Image-2.5, as part of a series of next-generation AI models. But how does it compare to the gold standard in image reproduction, Google’s Nano Banana?
When you think of Microsoft and its AI offerings like Copilot, you probably don’t think of creative work. The capabilities of these two new models — the standard 2.5 and the fast Flash one — are designed to change your mind.
Read more: Microsoft Build: All the Biggest AI Copilot, Agent and Chip News From San Francisco
“They give you intuitive programming with incredible control and consistency,” said Microsoft AI chief Mustafa Suleyman during Tuesday’s Build keynote. “Flash is here for high-performance workloads, while 2.5 gives you that high-fidelity and professional-level performance.”
Accurate editing of AI images is essential for professional or business work. This is an example that Microsoft shared of how its new model can adjust images.
These two new AI models were introduced by Microsoft at its developer conference Build on Tuesday. It also released its first “thinking” model, called MAI-Thinking-1. (The thinking model is designed to spend more time and make more iterations on the answer, ideally to produce better or to solve a more difficult question.) New generations of its voice and text models are also presented, as well as a code model optimized to work on GitHub, also owned by Microsoft. We also got an in-depth look at how Microsoft sees the future of computing, and — spoiler — it’s all about agent AI.
The new graphics model is available now in PowerPoint and in its business marketplace, Foundry, and is rolling out now to OneDrive.
This is a new way for Microsoft AI graphics models pile up the competition.
Which is better, MAI-Image-2.5 or Nano Banana 2?
When it comes to AI image processing, Google’s Nano Banana has dominated the creative AI space since its launch in 2025. For good reason — its power is the best in the industry, as it is powerful a flood of slop and deepfakes. That’s why it was surprising to hear Microsoft announce that its new graphics model has beaten the Nano Banana in one important way.
Microsoft’s MAI-Image-2.5 is better at image editing than Nano Banana 2, according to the ratings shown on the Arena AI leaderboard, a popular benchmark for AI models. But Microsoft only gets the silver medal –OpenAI’s GPT-Image-2 currently holds the top spot.
The usefulness of an AI model is not solely defined by any single benchmark, however. Accessibility plays a big role. One simple question will determine which will be more useful to you: Do you use PowerPoint or Google Slides? That ease of access to Microsoft’s and Google’s AI models will influence which one you end up using. If you’re using your AI images for commercial use, you’ll want to make sure you have the rights, which may depend on whether you’re using the AI for a business or personal purpose.
