Microsoft competes with Sora 2 and Nanobana with its own generative AI

Microsoft has unveiled its new AI MAI-Image 1 to enter the image generation race alongside Sora 2 from ChatGPT And NanoBanana of Gemini. However, where these competitors were banking on total freedom of transformation (in a Ghibli style or why not Disney), Microsoft wants to specialize in photorealism.

Microsoft enters the AI ​​race

The group’s new AI wants to stand out from the crowd. MAY-Image 1 “excels in generating photorealistic images“, assures Microsoft in a press release. More precisely in lighting or even landscapes. The company even boasted that it could avoid repetitive generations.

Microsoft says it designed MAI-Image-1 to provide real value to creators, avoiding stereotypical or repetitive images. To achieve this, the company relied on a more demanding selection of data and on tests inspired by concrete creative situations, while integrating feedback from professionals in the sector.

According to Microsoft, MAI-Image-1 stands out from larger and slower models thanks to its execution speed and image quality. The group emphasizes that this combination allows users to create, test and refine their creations more quickly, before reworking them in other tools for an optimized final result.

Microsoft AI for Microsoft

Microsoft is introducing MAI-Image-1 as a model offering more flexibility, visual diversity and real-world value for creators, and the first step in the company’s AI journey. However, the second issue for Microsoft is taking independence from OpenAI. Until then, the group used GPT-4o and DALL-E 3 models for services such as Bing Image Creator.

The MAI-Image-1 model is not yet available in Copilot or Bing Image Creator, but it is already making waves: it rose to 9th place in the LMArena ranking, a platform that compares AI performance according to user votes.

Want to save even more? Discover our promo codes selected for you.