TRENDING
Google Birthday 2026: How 28 Years of Search Led to the AI EraMuse AI Referral Code: Get 1 Billion Muse TokensClaude AI Found a New Enzyme System. Here's What Scientists KnowGoogle Just Gave Gemini a Face — Gemini 3.8 Live Avatar ExplainedGoogle's AI Agents Are Teaming Up to Make Longer VideosAustralia Investigates If OpenAI's AI Agent Broke the LawMeta Muse AI Glasses Explained: What's Real Right NowOpenAI's Agent Broke Into Australia's Medicare Site On Its OwnClaude Code Cloud Sessions: Claim Your $100 or $250 CreditAI Price War: Claude Opus 5.5 vs GPT-6 Sol and Luna ExplainedBristol Artists Criticize AI-Generated Mural After Visual ErrorsMeta Muse Zero-Day Explained: Can the AI Agent Be Hijacked?Claude Opus 5.5 Explained: Anthropic's New ModelJev AI Explained: The Decision Model That Returns Structured ChoicesAbhyas AI Explained: AI-Powered JEE & NEET Prep Platform

Microsoft's New AI Transcribes Audio for 10 Cents an Hour

Microsoft announced MAI-Transcribe-2 on September 3, 2026 — a speech-to-text model it says beats OpenAI, Google, and ElevenLabs on speed, accuracy, and price. It's built for developers, not a consumer app, but it's worth understanding if you rely on any tool that transcribes audio, since products built on it could show up in things you already use.

⚡ Quick facts

  • Speed: Up to 10x faster than GPT-Transcribe, 7x faster than ElevenLabs Scribe v2
  • Accuracy: 5.2% average word-error rate across 60 languages (FLEURS benchmark)
  • Price: $0.10/hour — an introductory rate lasting only through end of 2026
  • Access: Developers only, via Microsoft Foundry, MAI Playground, or OpenRouter
Advertisement

What it actually does

MAI-Transcribe-2 converts audio into text, with a set of features aimed at real-world use rather than just basic transcription:

How it compares to the competition

Microsoft's own benchmarks claim MAI-Transcribe-2 is up to 10x faster than OpenAI's GPT-Transcribe, 7x faster than ElevenLabs' Scribe v2, and 5x faster than Google's Gemini 3.5 Transcribe. On accuracy, it topped the FLEURS benchmark across 60 languages with a 5.2% average word-error rate — meaning it made fewer mistakes than rival models in that test, according to Microsoft.

Pricing: cheap, but only for now

At launch, MAI-Transcribe-2 costs $0.10 per hour of audio — but this is explicitly a limited-time introductory rate running only through the end of 2026. If you're building something on it, don't assume this price is permanent.

How to actually access it

This isn't an app you download. It's available to developers through Microsoft Foundry (via Azure Speech), the MAI Playground, and OpenRouter. If you're not building software yourself, you likely won't interact with it directly — but keep an eye out for it showing up under the hood in Microsoft products like Teams captions or Word dictation.

Frequently asked questions

What is MAI-Transcribe-2?

Microsoft's new speech-to-text model, announced September 3, 2026.

Is it actually better than competitors?

By Microsoft's own benchmarks, yes — faster and more accurate than OpenAI, Google, and ElevenLabs' transcription models.

How much does it cost?

$0.10/hour of audio, a limited-time price through end of 2026.

Can I use it directly?

Only if you're a developer, via Microsoft Foundry, MAI Playground, or OpenRouter.

Related articles

Comments