Microsoft MAI-Transcribe-2 cuts speech-to-text 72% to $0.10 per hour
Microsoft AI released MAI-Transcribe-2 on September 3 at $0.10 per audio hour, with a 5.2% word error rate on FLEURS and 10x faster than OpenAI GPT-Transcribe per Artificial Analysis.
Microsoft AI released MAI-Transcribe-2 on September 3 at $0.10 per audio hour, a 72% cut from the previous $0.36 rate set in April. The limited-time price runs through December 31, 2026.
The model adds speaker diarization, word-level timestamps, keyword biasing, automatic language identification, verbatim and clean transcription styles, code-switching support for Hinglish and Spanglish, and noise robustness. Language coverage expands from 43 to 60.
Microsoft claims MAI-Transcribe-2 ranks first on the FLEURS multilingual benchmark with a 5.2% average word error rate, and second on the Artificial Analysis word-error-rate leaderboard at 2.0%. On Artificial Analysis evaluations, the model posts a speed factor of 410.7, transcribing one hour of audio in under 9 seconds, 10x faster than OpenAI GPT-Transcribe, 7x faster than ElevenLabs Scribe v2, and 5x faster than Gemini 3.5 Transcribe. The model is available through Microsoft Foundry, the MAI Playground, and OpenRouter. This is the third transcription model from the Mustafa Suleyman-led Microsoft AI team in five months.