The model has already claimed a position within the top 10 of LMArena.
The model has already claimed a position within the top 10 of LMArena.
Microsoft AI has just rolled out its inaugural text-to-image generator, MAI-Image-1, crafted and created in-house. The technology titan, which recently unveiled its first in-house Microsoft AI models, refers to the novel image generator as “the next step on our journey.”
Microsoft asserts it consulted with creative professionals to prevent “repetitive or generically-stylized outputs.” MAI-Image-1 “excels” at creating photorealistic visuals, including lightning, landscapes, and beyond, according to the company. Furthermore, it can process requests and generate images more swiftly than “larger, slower models.” The model has already claimed a position within the top 10 of LMArena, the AI benchmark platform where users compare outputs from various systems and vote for their favorites.
MAI-Image-1 joins Microsoft’s other AI offerings, including the voice generator MAI-Voice-1 AI and the chatbot MAI-1-preview. The company was an early investor in OpenAI, but the relationship between the two entities has become increasingly intricate. Recently, Microsoft has started to incorporate Anthropic’s AI models into certain features of Microsoft 365, while also making “significant investments” in training its AI models like MAI-Image-1.
We haven’t yet had the opportunity to evaluate Microsoft’s new image generator, but once we do, we will provide feedback on its safety measures. The company states it is dedicated to “ensuring safe and responsible results.”
