Microsoft is finally joining the AI image generation craze with its newly announced MAI-Image-1. Though it may not be the company’s first stint into creating a text-to-image generation program, it is the first one to ever be developed in-house.
What differentiates this model compared to the competition, according to Microsoft, is that it has been trained to specifically avoid repetitive or “generically-stylised” results. The tech giant claims that MAI-Image-1 is really good at creating photorealistic images and can generate images with natural lighting and landscapes.
Besides the bump in generation quality, Microsoft claims that it’s quite fast when it comes to producing results. The company especially highlighted that, with this model, “users can get their ideas on screen faster, iterate through them quickly, and then transfer their work to other tools to continue refining.”
The tech giant also announced that the AI model debuted in the top 10 text-to-image models in LMArena. And the company is right, as the MAI-Image-1 is currently seated at 9th place at the time of writing. For those unaware, LMArena is an open-source platform that uses blind, head-to-head comparisons to assess and rank large language models (LLMs).
Microsoft concluded the blog post by saying that the model will be available in Copilot and Bing Image Creator “very soon”. In the meantime, the tech giant is inviting users to join LMArena to see how MAI-Image-1 compares to the rest of the competition.
(Source: Microsoft [Blog])
