Microsoft AI cuts MAI-Transcribe-2 to $0.10 per audio hour

Microsoft AI dropped its speech-to-text service to ten cents per audio hour on Thursday, roughly a 72% cut from the $0.36 it was charging five months ago. That price for MAI-Transcribe-2, reported by VentureBeat, undercuts OpenAI, Google and ElevenLabs on the same workload. An enterprise processing 100,000 hours of audio a year would see its bill fall from $36,000 to $10,000.

The model handles 60 languages, up from 43 in June's MAI-Transcribe-1.5 and 25 in April's original release. Microsoft says it ranks first on the FLEURS multilingual benchmark with an average 5.2% word error rate, and sits number two on the Artificial Analysis word-error-rate leaderboard. Per that same third-party leaderboard, the company claims MAI-Transcribe-2 runs "10 times faster than OpenAI's GPT-Transcribe, seven times faster than ElevenLabs' Scribe v2, five times faster than Google's Gemini 3.5 Transcribe." Those figures are vendor-cited, not independently reproduced here.

Alongside faster and cheaper, the release adds speaker diarization, word-level timestamps, keyword biasing for domain jargon, automatic language ID, and code-switching for mixed audio like Hinglish and Spanglish. VentureBeat is blunt about why diarization matters: "Speaker diarization sorts out who said what in a multi-person recording, which is the difference between a wall of text and a usable meeting transcript." It lands the same day VibeVoice-ASR shipped open weights for streaming speaker-attributed recognition, one of 40 voice AI stories we've tracked in the last 90 days.

The release is the latest MAI drop under Mustafa Suleyman since Microsoft restructured its OpenAI partnership last October and began using its own MAI models inside Word and Excel in July. VentureBeat frames the strategy plainly: "This is not an attempt to build one model that beats GPT or Gemini at everything." It sits inside a busy run of Microsoft coverage on our tracker this week.