Skip to main content
Current modelMicrosoft AI

MAI-Transcribe-2 benchmarks, pricing and specs

MAI-Transcribe-2 is a Microsoft AI model released on September 3, 2026 that takes audio input and produces text. It costs $0.001667 per audio minute. It scores 2.04 % WER on Artificial Analysis AA-WER v2, #1 of 6 current models (Artificial Analysis, measured October 13, 2026).

Per audio minute
$0.001667
Artificial Analysis AA-WER v2
2.04 % WER · #1/6

Data verified October 13, 2026 · Dataset 2026-10-13

Share this

Key facts

Provider
Microsoft AI
Model family
MAI-Transcribe
Release date
September 3, 2026
Status
Current
Access
API only
API model id
MAI-Transcribe-2
Context window
Not reported
Max output
Not reported
Input
audio
Output
text
Task families
audio
Declared capabilities
speech to text, diarization

MAI-Transcribe-2 pricing

It is tied for the cheapest of 12 current audio models priced per audio minute.

UnitPrice (USD)TierAppliesSource
1M input tokensNot reportedstandardCurrent
1M output tokensNot reportedstandardCurrent
1M input tokensNot reportedcachedCurrent
audio minute$0.001667standardUntil December 31, 2026

Independent benchmark results

Scores come from independent evaluators that ran the model themselves, pinned to a dated snapshot. Rank is among current models in the same snapshot. Microsoft AI's own launch figures are listed under strengths with the publisher named, because they are not measured the same way.

BenchmarkScoreRankMeasuredSource
Artificial Analysis AA-WER v22.04 % WER#1 of 6October 13, 2026Artificial Analysis · 2026-10-13

Strengths, cautions and best uses

Strengths

  • Microsoft reports first place on FLEURS across 60 languages with a 5.2% average word error rate
  • Diarization, word timestamps, keyword biasing, language identification and code-switching
  • $0.10 per hour of audio at launch

Cautions

  • Azure Speech lists it as a public preview with no SLA, not recommended for production
  • The $0.10 per hour price is a limited-time offer until 2026-12-31, and the later price is not published

Best for

  • Low-cost multilingual batch transcription
  • Meeting and call transcripts with speaker labels

Alternatives to MAI-Transcribe-2

The closest current audiomodels by price. The cheapest model that scores as well or better is listed first; failing that, the cheapest one within 3% of the benchmark's scale.

Compare MAI-Transcribe-2 with

MAI-Transcribe-2 FAQ

How much does MAI-Transcribe-2 cost?

MAI-Transcribe-2 costs $0.001667 per audio minute at the standard tier. It is tied for the cheapest of 12 current audio models priced per audio minute. Source: Microsoft AI, verified October 13, 2026.

How does MAI-Transcribe-2 score on independent benchmarks?

MAI-Transcribe-2 scores 2.04 % WER on Artificial Analysis AA-WER v2, #1 of 6 current models (Artificial Analysis, measured October 13, 2026). These are independent evaluations, not Microsoft AI's own launch figures.

Is MAI-Transcribe-2 open source or API only?

MAI-Transcribe-2 is available through Microsoft AI's API only; its weights are not published.

What are the best alternatives to MAI-Transcribe-2?

The closest current alternatives by price and measured performance are Grok Voice Transcribe 2.0 (same price per audio minute, $0.001667), Scribe v2 (2.2x the price per audio minute, $0.003667 vs $0.001667), GPT Transcribe (2.7x the price per audio minute, $0.0045 vs $0.001667) and Gemini 3.5 Transcribe (3.0x the price per audio minute, $0.005 vs $0.001667).

Sources

Cite this page

Van Data Team. "MAI-Transcribe-2 Benchmarks, Pricing and Specs." AI Model Benchmark Dataset 2026-10-13. https://vandatateam.com/ai-model-benchmarks/mai-transcribe-2

Reuse the data: JSON · CSV · or explore every model in the AI model benchmark comparison.

Production evaluation

Deciding whether MAI-Transcribe-2 fits your workload?

Public benchmarks narrow the shortlist. We test the finalists on your own tasks, data and cost limits, then build the routing and review controls to run the winner in production.

  • Evaluation cases and acceptance thresholds tied to real work
  • Model routing, fallbacks and cost controls