Microsoft Launches Three Rival AI Models

Microsoft AI unveiled three foundational models: MAI-Transcribe-1 for speech-to-text in 25 languages, MAI-Voice-1 for audio generation, and MAI-Image-2 for image output. All three are now available on Microsoft Foundry, developed by the MAI Superintelligence team led by CEO Mustafa Suleyman. Microsoft positions the models as cheaper alternatives to Google and OpenAI offerings, with transcription starting at $0.36/hour. Despite building its own AI stack, Microsoft reaffirmed its $13 billion OpenAI partnership, with a recent renegotiation enabling the company to pursue independent superintelligence research.