Audio · Speech-to-text · Released Sep 2026
Microsoft MAI-Transcribe 2
Microsoft AI's multilingual transcription model, which covers 60 languages with automatic language detection and returns plain text.
From Microsoft, MAI-Transcribe 2 is a cloud speech-to-text model. It transcribes spoken audio into written text. It handles multiple spoken languages. It runs through OpenRouter using your own API key, from about $0.0017 per minute of audio.
Specs
- Released
- this month (Sep 2026)
- Pricing
- $0.0017 per minute of audio
- Type
- Speech-to-text
- Languages
- Multilingual
About the creator
Microsoft
Microsoft Research builds the open-weight Phi family of small language models — efficient, reasoning-focused models that punch above their parameter count.
azure.microsoft.com/products/phi ↗More from Microsoft
Phi-4 14B
TextLocal
Microsoft's 14B state-of-the-art open model with strong reasoning, math, and code.
Phi-4 Mini 3.8B
TextLocal
Microsoft's compact 3.8B model with a 128K context and native function calling.
Phi-4 Reasoning 14B
TextLocal
Microsoft's 14B reasoning model fine-tuned for chain-of-thought on math, science, and code.