New Delhi: Microsoft has unveiled its first fully in-house image generation model called MAI-Image-1, marking its formal entry into the visual AI race dominated by Google’s Gemini-2.5 Flash “Nano Banana” and OpenAI’s GPT-Image-1. The company said the model, now available for public testing on LMArena, will soon roll out to Copilot and Bing Image Creator.

The new model secured the #9 spot on the global text-to-image leaderboard at LMArena, a platform where users compare anonymous AI models and vote for the best outputs. This debut positions Microsoft among the top 10 image generation systems globally, a notable entry for a model built entirely in-house.

Microsoft’s first homegrown image model

MAI-Image-1 was trained by Microsoft’s internal AI team with a focus on creative precision and

See Full Page