Microsoft launches MAI-Image-1 for Bing, Copilot

Microsoft has rolled out MAI-Image-1, its first in-house text-to-image model, making it available inside Bing Image Creator and Copilot Audio Expressions. The company first announced the model in October; Microsoft AI chief Mustafa Suleyman said in a post on X that MAI-Image-1 will be “coming soon” to the EU, but it’s not available there yet.

Microsoft and Suleyman highlight the model’s strengths in photorealistic imagery — especially food and nature scenes — and in rendering nuanced lighting (bounce light, reflections) and other fine details. Microsoft’s blog emphasizes MAI-Image-1’s combination of speed and quality compared with many larger, slower models, saying that enables faster iteration and easier handoff to other tools.

MAI-Image-1 will also produce artwork to accompany AI-generated audio stories in Copilot Audio Expressions’ “story mode.” The launch follows Microsoft’s earlier in-house releases (MAI-Voice-1 for speech and MAI-1-preview for text). Microsoft still uses OpenAI models in some places — Bing Image Creator’s model list includes DALL·E 3 and GPT-4o from OpenAI — and the company says Copilot is transitioning to GPT-5 while also offering Anthropic’s Claude models as options.

Source