MICROSOFT
Microsoft just built its own AI artist
Microsoft has launched its first in-house text-to-image model, MAI-Image-1, marking a new step in its growing AI lineup.
The model was built with help from creative professionals to avoid the “generic” or repetitive look seen in many AI image tools.
Microsoft says MAI-Image-1 performs well with realistic visuals like lightning and landscapes and can generate images faster than larger models.
It’s already in the top 10 on LMArena, a site where people compare AI-generated images.
MAI-Image-1 joins other Microsoft tools like MAI-Voice-1 and MAI-1-preview, its chatbot.
Three things to know:
MAI-Image-1 is Microsoft’s first homegrown text-to-image model.
It’s already performing well in public benchmarks.
Microsoft’s AI focus is expanding beyond OpenAI partnerships.
Built, not borrowed
While Microsoft was one of OpenAI’s first major investors, the two companies’ relationship has become more complicated.
Reports say Microsoft plans to use Anthropic’s models for some Microsoft 365 features.
The company hasn’t shared details on MAI-Image-1’s safety features yet but says it’s focused on “safe and responsible outcomes.”
Can Microsoft finally break my Midjourney addiction? - MV