Microsoft launches MAI Image 2 model for Copilot and Bing Image Creation

Microsoft has introduced its latest AI image generation model, MAI Image 2, promising to deliver more lifelike images and improved readability in generated text. This new model was developed through collaboration with photographers, designers, and visual storytellers, refining its capabilities to enhance realism. The MAI Image 2 aims to produce images characterized by natural lighting, realistic skin tones, and environments that appear authentic. This focus on realism reduces the need for extensive post-processing, thereby allowing users to save valuable time. As highlighted in a recent blog post, Microsoft emphasizes that "MAI-Image-2 is designed for creatives seeking images that feel real, with natural light and lived-in settings. Creatives can focus more on their craft rather than fixing details in post-production." In addition to its photorealistic improvements, the MAI Image 2 has made significant strides in generating in-image text, providing accurate text for various applications like infographics and diagrams. Users can currently preview this model on the MAI Playground, and it is gradually being integrated into Copilot and Bing Image Creator. Microsoft is committed to facilitating seamless workflows for designers, marketers, and developers, enabling them to generate high-quality visuals directly within their preferred applications. API access is currently limited to select Microsoft customers, with broader availability expected through Microsoft Foundry for developers in the near future. In a noteworthy achievement, the MAI Image 2 has secured the #3 position among AI labs for text-to-image generation globally, according to the Arena.ai leaderboard. However, Microsoft faces stiff competition from other leading AI image models, including Google’s Nano Bana 2, OpenAI’s DALL-E 3, and Midjourney 6.1, all of which are pushing the boundaries of what's possible in AI-generated imagery.