By No Code MBA · Facts checked
- Provider
- Microsoft
- Announcement date
- Status at last check
- Generally available
- Model family
- MAI
- Preceding model
- MAI-Transcribe-2
- Capabilities
- Audio
What changed
MAI-Transcribe-2-Streaming is Microsoft's first streaming speech-to-text model, covering 60 languages with continuous language detection and first partial results after just over 100 ms of audio; Microsoft says it ranks first for accuracy on Artificial Analysis for final and partial transcripts. MAI-Voice-2.1 and MAI-Voice-2.1-Flash are text-to-speech models for 23 languages with voice cloning from a few seconds of audio. Microsoft says Flash has about 150 ms end-to-end latency.
Access and availability
All three models are available in Microsoft Foundry, with Azure Voice Live, MAI Playground and Vercel also listed; the voice models are on OpenRouter, and LiveKit support is listed as coming soon.
Official access documentationPricing
MAI-Transcribe-2-Streaming $0.54 per hour of audio (introductory, through end of 2026); MAI-Voice-2.1 $22 per million characters; MAI-Voice-2.1-Flash $15 per million characters
Official pricingWhat you can build
Prototype a voice-agent loop using representative call audio, then compare transcription latency and short-utterance accuracy alongside response quality and voice consistency across the languages you need.
Limitations
No WER figures are published, and speed comparisons come from Microsoft's internal evaluations. Flash generates up to 45 seconds of audio per request, and the transcription price is an introductory rate through the end of 2026.