MAI-Transcribe-2-Streaming

Overview

MAI-Transcribe-2-Streaming is Microsoft‘s first-party real-time streaming automatic speech recognition (ASR) model, released on 2026-10-02 across 60 languages with ~100ms first-partial latency. Microsoft reports it ranks first on the Artificial Analysis streaming-transcription leaderboard. Shipped externally through Microsoft Foundry in public preview (no SLA, not production-recommended) at $0.54/hr audio introductory pricing through year-end. Framed by the corpus as the second iteration of Microsoft’s post-renegotiation OpenAI-independence stack — additive to Azure OpenAI’s Whisper, not a mandated swap.

Timeline

• 2026-10-03-AI-Digest — MAI-Transcribe-2-Streaming launches on 2026-10-02 as a Microsoft-native streaming ASR alternative to Whisper on Foundry — 60 languages, ~100ms first-partial latency, intro pricing $0.54/hr through year-end. Microsoft claims first place on Artificial Analysis’s streaming-transcription leaderboard. Corpus framing: second iteration of Microsoft's post-renegotiation independence stack, not Microsoft replaces OpenAI's voice stack — MAI-Transcribe-1 already beat Whisper-large-v3 across all 25 FLEURS languages at 50% lower GPU cost (April 2026); today’s release is the hardening pass, not the strategic turn. Shipped in public preview on Foundry — no SLA, additive to Azure OpenAI’s Whisper rather than a mandated replacement.

Key Developments

1. MAI-Transcribe-2-Streaming public preview on Foundry (October 2, 2026): Real-time streaming ASR across 60 languages, ~100ms first-partial latency, intro pricing $0.54/hr through year-end. Launched alongside MAI-Voice-2.1. Microsoft reports top rank on Artificial Analysis streaming-transcription leaderboard. Public preview — no SLA, not production-recommended; sits side-by-side with Azure OpenAI’s Whisper on Foundry rather than replacing it.

Related

See also: Microsoft, MAI-Voice-2.1, MOC - Major Companies, MOC - Developer Tools.