Grok Voice Transcribe 2.0
xAI releases Grok Voice Transcribe 2.0, claiming top streaming transcription accuracy, twice v1.0's accuracy at unchanged pricing.
xAI launched Grok Voice Transcribe 2.0, a speech-to-text model built on the audio foundation model behind Grok Voice, claiming it ranks first in accuracy among 32 streaming models on the Artificial Analysis leaderboard. Internal evaluations show short-phrase word error rate dropping from 20.6% to 6.8%, with multilingual transcription the largest improvement over version 1.0. Features include batch and streaming modes, word-level timestamps, speaker diarization, up to 8-channel transcription, and key-term biasing. Pricing is unchanged at $0.10 per hour batch and $0.20 per hour streaming; version 2.0 becomes the Speech-to-Text API default while 1.0 will be deprecated, and Atlassian is adopting it for Loom transcription.
- Ranks first for accuracy among 32 streaming models on the Artificial Analysis leaderboard.
- Short-phrase word error rate falls from 20.6% to 6.8%; multilingual accuracy is the biggest gain.
- Pricing unchanged: $0.10/hour batch, $0.20/hour streaming, with diarization and timestamps included.
- Powers Grok Voice in Tesla vehicles and tens of thousands of daily customer-support calls; adopted by Atlassian Loom.
- Version 1.0 will be deprecated in coming weeks as 2.0 becomes the API default.
From the original article
3 min read at x.ai
Today we're releasing Grok Voice Transcribe 2. 0, our latest speech-to-text model. Across our real-world evaluations, Grok Voice Transcribe 2.
By vertigoruntime · x.ai
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.