ModelCards

Grok Voice Transcribe 2.0

Audio modelCard summarized

SpaceXSI's newest speech-to-text model, with unparalleled accuracy and cost effectiveness.

Source: Introducing Grok Voice Transcribe 2.0 (launch post) and Speech to Text docs; no model card published

Use it via API

$0.10 / hour
Code exampleCompare

Where to call it

  • xAI APIgrok-voice-transcribe-2.0

Supports

  • Streaming
Full feature support

Good for

  • Telephony and customer-support audio (8 kHz)
  • Spoken credentials: phone numbers, emails, addresses
  • Short multilingual voice commands

Watch out for

Overview

xAI · 01Transcribe, front

Transcribe 2.0

SpaceXSI's newest speech-to-text model, with unparalleled accuracy and cost effectiveness.

Typical price
$0.10 / hour
Batch (REST)
Max length
—
Languages
38+ languages, auto-detected, with mid-recording switches
Modes
speech-to-text, streaming
Released
18 Sep 2026

Grok Voice Transcribe 2.0 is xAI's speech-to-text model for recorded files and real-time streams, built on the audio foundation model behind Grok Voice. xAI says it is twice as accurate as Grok Voice Transcribe 1.0 at the same price, ranks first for accuracy among 32 streaming models on the Artificial Analysis leaderboard, and makes its largest gains on multilingual audio. It is the default model in the Speech to Text API. launch post; Speech to Text docs; release notes

grok-voice-transcribe-2.0
Typical price
$0.10 / hour
Languages
38+ languages, auto-detected, with mid-recording switches
Weights
Closed
Released
18 Sep 2026
  • speech-to-text
  • streaming
  • speaker diarization
  • word-level timestamps
  • multichannel (up to 8)
  • key term biasing
  • Smart Turn end-of-turn detection