Updated

Grok Voice Transcribe 2.0

xAI released Grok Voice Transcribe 2.0, a speech-to-text model available through its Speech-to-Text API for batch and streaming transcription. Company-listed pricing is unchanged from version 1.0: $0.10 per hour of audio for batch and $0.20 per hour for streaming.

By No Code MBA · Facts checked

Provider
xAI
Announcement date
Status at last check
Generally available
Model family
Grok Voice
Preceding model
Grok Voice Transcribe 1.0
Capabilities
Audio, Text

What changed

Grok Voice Transcribe 2.0 is xAI's speech-to-text model built on the Grok Voice audio foundation model, with automatic language detection and mid-recording language switching across 38+ languages. xAI says it is twice as accurate as version 1.0 at the same price, with the largest gains in multilingual audio, and reports short-phrase word error rate fell from 20.6% to 6.8%. xAI also says it ranks first for accuracy among 32 streaming models on the Artificial Analysis leaderboard.

Access and availability

Use model grok-voice-transcribe-2.0 (the default when model is omitted) via POST https://api.x.ai/v1/stt for batch or wss://api.x.ai/v1/stt for streaming. Diarization, timestamps and key terms are included.

Official access documentation

Pricing

$0.10 per hour of audio for batch and $0.20 per hour for streaming

Official pricing

What you can build

Send a noisy support call or meeting recording through the Speech-to-Text API and compare speaker labels, timestamps, and formatted emails or phone numbers against your current transcriber.

Limitations

Files are capped at 500 MB, and text formatting (inverse text normalization) works in only 11 languages. Most accuracy results are xAI's internal tests, and version 1.0 will be deprecated in the coming weeks.

Sources and release history