Key Points
- Microsoft has unveiled its first in-house AI image generator, called MAI-Image-1.
- The move is part of a broader strategy to reduce its reliance on its partner, OpenAI.
- The new model is said to be particularly good at creating photorealistic images.
- It will soon be integrated into Microsoft’s Copilot assistant and Bing Image Creator.
Microsoft is continuing its push to develop its own AI models, announcing its first internally developed image generator on Tuesday. The new model, called MAI-Image-1, is a direct move to decrease the company’s reliance on its high-profile partner, OpenAI.
According to Microsoft, MAI-Image-1 is particularly good at creating photorealistic images, especially when it comes to natural lighting and landscapes. The model is currently being tested, but Microsoft says it will be coming to its Copilot assistant and Bing Image Creator “very soon.”
This is the latest in a series of in-house AI releases from the tech giant. Over the summer, Microsoft unveiled its first two homegrown models, one for voice and one for language. At the time, the head of Microsoft’s AI division, Mustafa Suleyman, said the company has an “enormous five-year roadmap” for its own AI development.
For years, Microsoft’s AI strategy has been closely tied to its multi-billion-dollar investment in OpenAI. But these recent moves show that the company is also building a powerful, independent AI capability. By developing its own models, Microsoft gains more control, can tailor them to its specific products, and reduces its dependence on a single, powerful partner.
The release of MAI-Image-1 puts Microsoft in more direct competition with other major players in the AI image generation space, including OpenAI’s DALL-E and Midjourney.
