MAI-Transcribe-2-Streaming
Microsoft AI's first real-time streaming speech-to-text model, ranked #1 for transcript accuracy on Artificial Analysis across 60 languages.
Overview
Microsoft AI released MAI-Transcribe-2-Streaming on October 1, 2026, its first real-time transcription model, which transcribes words as they arrive rather than after a recording finishes. Microsoft reports a 2.5% word error rate with a final transcript delivered about 0.13 seconds after speech ends, across 60 languages with automatic language detection, and Artificial Analysis ranks it #1 of 38 models for final and first-partial transcript accuracy. It is priced at $0.54 per hour of audio as an introductory rate through the end of 2026.
This is a sourced directory entry. Features and capabilities are reported in the linked sources; this entry does not document independent Aymorphic testing.
Key features
- True streaming transcription: partial transcripts as speech arrives, not just after the fact
- 2.5% word error rate; final transcript ~0.13s after speech ends
- 60 languages with automatic language detection
- Ranked #1 of 38 models on Artificial Analysis for transcript accuracy
- Introductory pricing of $0.54 per audio hour through end of 2026
Use cases
- Live captioning and real-time meeting transcription
- Voice agents needing low-latency speech-to-text
- Multilingual customer support call transcription
Pricing
$0.54 per hour of audio, introductory rate through end of 2026.
Pricing reflects the source record and may change. Confirm current terms with the provider.
Dates and sources
- Released
- 1 Oct 2026
- Record updated
- 4 Oct 2026
- Record last checked
- 4 Oct 2026