LLM Reference
Microsoft Foundry

Using MAI-Transcribe-2 on Microsoft Foundry

Implementation guide · MAI · Microsoft AI

Serverless

Microsoft Foundry exposes MAI-Transcribe-2 through model ID MAI-Transcribe-2. Use the setup steps, sourced pricing, capabilities, and official provider links below to validate this route before deployment.

Last refreshed 2026-09-03. Next refresh: weekly.

Quick Start

  1. 1
    Create an account at Microsoft Foundry and generate an API key.
  2. 2
    Use the Microsoft Foundry SDK or REST API to call MAI-Transcribe-2 — see the documentation for request format.

Code Examples

See Microsoft Foundry documentation for integration details.

Pricing on Microsoft Foundry

Capabilities

MultimodalAudio

About MAI-Transcribe-2

MAI-Transcribe-2 (Azure Speech enhancedMode.model id MAI-Transcribe-2) is Microsoft AI's third-generation speech-to-text / ASR model, announced September 3, 2026. Specialized transcription API, not a chat LLM. First-party: 60 languages; speaker diarization; word-level timestamps; keyword biasing; verbatim/clean transcription styles; code switching; automatic language identification; noise-robust real-world audio. Microsoft reports first place on FLEURS across 60 languages at 5.2% average WER, Artificial Analysis accuracy-latency Pareto leadership, and second place on the Artificial Analysis WER leaderboard. Available to demo via Microsoft Foundry, MAI Playground, and Open Router; Azure Speech docs mark MAI-Transcribe as public preview. Compare it for Speech recognition.

Model Specs

Released2026-09-03
ArchitectureAudio / Speech

Provider

Microsoft Foundry
Microsoft Foundry

Microsoft

Redmond, Washington, United States