Particle.news
Download on the App Store
10 ARTICLES
·
6h ago

Microsoft Launches MAI-Image-1, Its First In-House Text-to-Image Model

The release signals Microsoft’s push to build proprietary AI beyond OpenAI.

Image
Microsoft intros MAI-Image-1 for Copilot, Bing Image Creator
Cube with Microsoft logo on top of its office building on 8th Avenue and 42nd Street near Times Square in New York City.
In recent months, Google’s quirky “Nano Banana” feature, which lets users generate humorous and viral AI images, has taken social media by storm.

Overview

  • MAI-Image-1 debuted on LMArena’s leaderboard at No. 9 with a reported score of 1,096, behind Hunyuan-image-3.0 at No. 1 and Google’s Gemini-2.5-Flash “Nano Banana” at No. 2, with OpenAI’s entry at No. 7.
  • Microsoft says the model focuses on photorealism, handling lighting, reflections, and landscapes particularly well while generating images faster than larger, slower systems.
  • The company cites rigorous data selection and direct feedback from creative professionals to reduce repetitive, generic styling in outputs.
  • Public testing is available now on LMArena, and Microsoft says integrations with Copilot and Bing Image Creator are coming “very soon.”
  • The launch expands Microsoft’s MAI lineup and its in-house capabilities alongside select partner models, with promised safeguards in place and independent evaluations still pending.