Krux
March 25, 2026
Microsoft's New Image Model Finally Nails In-Image Text
Published: March 25, 2026 at 12:34 AM
Updated: March 25, 2026 at 12:34 AM
100-word summary
Microsoft just launched MAI-Image-2, its in-house text-to-image model built to solve one of AI's most annoying problems: text that looks like alphabet soup. The model generates readable text inside images, meaning you can now create infographics, posters, and slides without opening Photoshop afterward. It also delivers sharper photorealism and richer scenes, refined with input from actual photographers and designers. Available now in MAI Playground and rolling out to Copilot and Bing Image Creator, with API access live for select customers like WPP. This is Microsoft's bet that building its own image generator beats licensing someone else's.
What happened
Microsoft just launched MAI-Image-2, its in-house text-to-image model built to solve one of AI's most annoying problems: text that looks like alphabet soup. The model generates readable text inside images, meaning you can now create infographics, posters, and slides without opening Photoshop afterward. It also delivers sharper photorealism and richer scenes, refined with input from actual photographers and designers. Available now in MAI Playground and rolling out to Copilot and Bing Image Creator, with API access live for select customers like WPP.
Why it matters
This is Microsoft's bet that building its own image generator beats licensing someone else's.