Microsoft takes on AI rivals with three new foundational models

Microsoft AI, the tech giant's research lab led by CEO Mustafa Suleyman, has launched three new foundational AI models designed to compete directly with offerings from Google and OpenAI. The models—MAI-Transcribe-1, MAI-Voice-1, and MAI-Image-2—handle speech-to-text, audio generation, and video generation, respectively. A key selling point is their competitive pricing, with MAI-Transcribe-1 starting at $0.36 per hour and MAI-Voice-1 at $22 per 1 million characters, undercutting many rivals. The release marks a significant step in Microsoft's strategy to build its own multimodal AI stack, even as it maintains its multi-billion dollar partnership with OpenAI.

MAI-Transcribe-1 offers speech transcription across 25 languages and is reportedly 2.5 times faster than Microsoft's existing Azure Fast service. MAI-Voice-1 can generate 60 seconds of audio in just one second and allows for custom voice creation. MAI-Image-2, a video-generating model, was first previewed on the MAI Playground testing software in March. Developed by the MAI Superintelligence team formed in late 2025, these models are now available on Microsoft Foundry, with the transcription and voice models also accessible via MAI Playground. Suleyman emphasized the company's 'Humanist AI' approach, focusing on practical use and human-centric communication, and promised more models will follow in Foundry and Microsoft products.