Microsoft launches two new AI models for image and voice

Microsoft launched two new AI models on July 23: MAI-Image-2.5-Pro and MAI-Voice-2-Flash, now in public preview. MAI-Image-2.5-Pro is Microsoft's most accurate image model for generating visuals, editing details, and rendering text in images, controlled by natural language. MAI-Voice-2-Flash is optimized for high-frequency voice tasks, twice as fast as its predecessor with 32% lower cost, ideal for call centers and voice assistants.

These models are already in Microsoft products. Bing Image Creator uses MAI-Image-2.5 as its default image generator. In PowerPoint, it cuts GPU costs by up to 84% compared to GPT-Image-2. In OneDrive, it boosts save success rates by 26% and reduces latency by 25%. MAI-Voice-2-Flash powers Dynamics 365 Contact Center for enterprise customer service agents, cutting GPU costs by up to 89%.

Microsoft is also integrating its models into specialized fields. MAI-Transcribe-1.5 supports Dragon Copilot for medical voice solutions, handling 58 languages with 50% fewer transcription errors. The next-gen GB200 computing cluster is now running to support further model development.