Microsoft opens MAI speech and image models to developers

Microsoft added MAI-Transcribe-1, MAI-Voice-1, and MAI-Image-2 to Foundry.

Pricing: transcription $0.36 per hour; voice $22 per 1M characters; image generation $5 per 1M text-input tokens + $33 per 1M image-output tokens.

MAI-Transcribe-1 leads the FLEURS benchmark across the top 25 languages, supports WAV/MP3/FLAC, and is tuned for captions, meeting notes, call analysis, accessibility, and voice agents; real-time transcription, diarization, and context biasing are not yet available.

MAI-Voice-1 can produce 60 seconds of audio in one second and lets customers create a custom voice in Foundry from a few seconds of audio; it was featured in Copilot Daily, Podcasts, and Copilot Labs in 2025.

MAI-Image-2 targets creative and commercial work and launched in MAI Playground on March 19, 2026; MAI Playground testing remains US-only, and Satya Nadella tweeted about the changes on April 2, 2026.