Identity
- Status
- preview
- Developer ID
mai-transcribe- Released
- Sep 3, 2026
- Version
- MAI-Transcribe 2
- License
- Unknown
MicrosoftMAI-Transcribe 2
MAI-Transcribe 2 is Microsoft's preview speech-to-text model for multilingual transcription with speaker diarization and word-level timestamps.
- Token context
- N/A
- Inputs
- Audio
- Outputs
- Text
- Released
- Sep 3, 2026
Providers & Pricing
Estimate workload cost →| Provider | Price dimension | Amount | Unit | Observed | Evidence |
|---|---|---|---|---|---|
| Transcription | $0.10 | per hour | Sep 3, 2026 | Provider-reported |
At a Glance
Model Facts
Capacity
- Token context
- N/A
- Maximum output
- N/A
- Total parameters
- Unknown
- Active parameters
- Unknown
- Knowledge cutoff
- Not reported
Interface and Access
- Inputs
- Audio
- Outputs
- Text
- API available
- Yes
- Open weights
- No
- Self-hostable
- Unknown
- Reasoning
- No
- Vision
- No
- Tool calling
- No
- Capabilities
- diarization, multilingual, speaker-attribution, transcription, word-level-timestamps
This page represents one developer model product. Dated API snapshots, serving endpoints, pricing, regions, quantizations, and service tiers attach as versioned aliases or provider details and never create duplicate public model records.
Version Lineage
PredecessorUnknown
SuccessorsUnknown
Aliasesmai-transcribe, mai-transcribe-2, MAI-Transcribe-2
Explicit Unknowns
- Architecture
- Tokenizer
- Knowledge cutoff
Primary Evidence
Sources and Observation Date
Comparable Peers
Related Model Comparisons
| A | Pair | B | Context |
|---|---|---|---|
MAI-Transcribe 2Microsoft | vs | VibeVoice-ASR-Streaming-1.5BMicrosoft | developer peersaudio, text |
MAI-Transcribe 2Microsoft | vs | VibeVoice-ASR-Streaming-7BMicrosoft | developer peersaudio, text |
MAI-Transcribe 2Microsoft | vs | cross-developer peersaudio, text | |
MAI-Transcribe 2Microsoft | vs | cross-developer peersaudio, text | |
MAI-Transcribe 2Microsoft | vs |  Lyria 3.5Google DeepMind | cross-developer peersaudio, text |
MAI-Transcribe 2Microsoft | vs | Gemini 2.5 FlashGoogle DeepMind | cross-developer peerstext |
MAI-Transcribe 2Microsoft | vs | Gemini 2.5 Flash-LiteGoogle DeepMind | cross-developer peerstext |
MAI-Transcribe 2Microsoft | vs | Gemini 2.5 ProGoogle DeepMind | cross-developer peerstext |
MAI-Transcribe 2Microsoft | vs | Gemini 3 FlashGoogle DeepMind | cross-developer peerstext |
MAI-Transcribe 2Microsoft | vs |  MiniMax-Music3MiniMax | cross-developer peersaudio |
MAI-Transcribe 2Microsoft | vs |  stable-audio-3-mediumStability AI | cross-developer peersaudio |
Questions
MAI-Transcribe 2 FAQs
What is MAI-Transcribe 2?+
MAI-Transcribe 2 is Microsoft's preview speech-to-text model for multilingual transcription with speaker diarization and word-level timestamps. It is developed by Microsoft and its current sourced lifecycle status is preview.
What inputs and outputs does MAI-Transcribe 2 support?+
MAI-Transcribe 2's current record lists Audio as input and Text as output.
Is MAI-Transcribe 2 available through an API?+
MAI-Transcribe 2 is marked as API-available. The directly sourced provider records currently include Microsoft Foundry.
How much does MAI-Transcribe 2 cost through an API?+
No directly sourced token price is currently attached to MAI-Transcribe 2. Missing prices remain unknown rather than being inferred from another model or provider.
What capabilities and tasks does MAI-Transcribe 2 support?+
MAI-Transcribe 2's recorded capabilities are diarization, multilingual, speaker-attribution, transcription, and word-level-timestamps. Its supported tasks are audio and transcription.
Can MAI-Transcribe 2 be self-hosted?+
MAI-Transcribe 2 is not marked as an open-weight model. Self-hosting is not established by the current source, with no license recorded.
Does MAI-Transcribe 2 have related versions or aliases?+
No predecessor is recorded. No successor is recorded. Recorded aliases are mai-transcribe, mai-transcribe-2, and MAI-Transcribe-2.
Which models can MAI-Transcribe 2 be compared with?+
The current compatible peer set includes VibeVoice-ASR-Streaming-1.5B, VibeVoice-ASR-Streaming-7B, Granite Speech 5.0 TurboCTC, Granite Speech 5.0 TurboCTC NC, and Lyria 3.5. Compatibility requires shared sourced modalities and tasks; it does not imply that one model is better. Browse model comparisons →
How current is the Model Markets record for MAI-Transcribe 2?+
MAI-Transcribe 2 was last verified Sep 3, 2026 from Microsoft's primary source, “MAI-Transcribe-2: Highest quality transcription, at the fastest speed and lowest cost.” The evidence is classified as official fact.