ModelsMAI-Transcribe-2-Streaming: real-time transcription from Microsoft
Microsoft AI added live transcription in 60 languages and two new voice models, also available through Vercel AI Gateway.
Source of truthmicrosoft.ai/…/our-first-streaming-tran…Microsoft announced MAI-Transcribe-2-Streaming on October 1, 2026, a speech-to-text model that returns text in real time in 60 languages, with automatic language detection. According to the company, it produces its first words in just over 100 ms. It costs $0.54 per hour of audio, an introductory price through the end of the year. Two text-to-speech models came with it: MAI-Voice-2.1, at $22 per million characters, and MAI-Voice-2.1-Flash, at $15. All three are on Vercel AI Gateway with no extra charge from Vercel. Whether Greek is supported is not stated.
Other sources: vercel.com/…/microsoft-ai-models-are-no…
Related stories
Anthropic releases Claude Sonnet 5.5 at the same price as Sonnet 5
A new Sonnet with no price change: $2 per million input tokens and $10 per million output tokens. According to Anthropic, it is faster and cheaper per task.
Suno Speech: voice and music in one track, in beta for everyone
Suno sets speech over its own music from a description, now open in beta to everyone.