Last year, Microsoft announced its in-house image generation model, MAI-Image-1, which debuted at the #9 position in Arena's Image Arena, significantly lagging behind other AI labs. The company made the model available to Bing.com/create and Bing mobile app users.
In March, the Microsoft AI team introduced MAI-Image-2, its second-generation image generation model, featuring significant improvements with the ability to generate images with better natural light, accurate skin tones, and more. MAI-Image-2 debuted at the #3 position, trailing only Google's gemini-3.1-flash-image-preview and OpenAI's gpt-image-1.5-high-fidelity. This model was made available inside Copilot and Bing Image Creator. It was also made available for developers via API on Microsoft Foundry.

Today, the Microsoft AI team announced MAI-Image-2.5, its latest text-to-image model that is now ranked third on Arena’s latest text-to-image leaderboard. OpenAI's gpt-image-2 is currently leading the pack with a score of 1388.
According to Microsoft, this new MAI-Image-2.5 model performs better across a broad range of image styles. The model is designed to follow prompts more closely, render text more reliably, and produce more detailed and coherent images. Microsoft also says the model has stronger visual reasoning, allowing it to better understand objects, lighting, scale, scene structure, and spatial relationships.
Microsoft highlighted that this new model delivers the biggest improvement in text rendering, stylized illustrations, and commercial imagery, allowing users to generate better posters, packaging mockups, branding concepts, and product shots. The text in generated images will be sharper, the layouts will be more stable, and brand-focused visuals will come through with more polish.
Meet MAI-Image-2.5 - ranked third on the @arena text-to-image leaderboard. It's another great advance in quality. And with Build just a week away, there's much more to come from the @MicrosoftAI team. I can't wait. pic.twitter.com/11wx96Z04a
— Mustafa Suleyman (@mustafasuleyman) May 26, 2026
The new MAI-Image-2.5 model is available today on Arena for anyone to try. Like previous Microsoft AI models, this new image generation model will also be made available on the MAI Playground and Microsoft Foundry within the next two weeks.
4 Comments
Load the comments and join the conversation!
Read the comments, ask the editors questions, show respect and join the conversation.