Microsoft AI released MAI-Transcribe-2 on September 3, positioning it as the fastest, most accurate and cheapest speech recognition model available. The model tops the FLEURS benchmark across 60 languages with a 5.2% average word-error rate, beating OpenAI's GPT-Transcribe, ElevenLabs' Scribe v2 and Google's Gemini 3.5 Transcribe.
Microsoft says the model processes long-form audio up to 10 times faster than GPT-Transcribe, while undercutting rivals on price at just $0.10 per hour of audio through the end of 2026 — a steep drop from the $0.36 launch price of its predecessor. New features include speaker diarization, word-level timestamps, keyword biasing, and automatic code-switching for mixed-language conversations like Hinglish or Spanglish.
The model is available now in public preview through Microsoft Foundry, the MAI Playground and Open Router, intensifying competition in the fast-growing market for AI transcription tools.