xAI · Sunday, September 20, 2026
xAI releases Grok Voice Transcribe 2.0 speech-to-text model
The File desk · Sep 20, 2026, 11:41 AM UTC
Status
Confirmed (xAI news post, Sept. 18, 2026)
xAI released Grok Voice Transcribe 2.0 on Friday — a new speech-to-text model built on the audio system behind Grok Voice. The company says the model is twice as accurate as Transcribe 1.0 at the same price and ranks first for accuracy among 32 streaming models on the public Artificial Analysis leaderboard. It is trained for noisy, real-world audio, including phone calls and multilingual speech. Existing Speech-to-Text API customers get the upgrade without changing code. Pricing stays $0.10 per hour of audio for batch work and $0.20 per hour for streaming, with speaker labels, timestamps, and key-term biasing included. On one internal short-phrase test set, the company reports word error rate falling from 20.6 percent to 6.8 percent.
xAI released Grok Voice Transcribe 2.0 on Friday — a new speech-to-text model built on the audio system behind Grok Voice.
The company says the model is twice as accurate as Grok Voice Transcribe 1.0 at the same price.
On the public Artificial Analysis leaderboard, xAI says it ranks first for accuracy among 32 streaming models.
The company says it is trained for noisy, real-world audio, including phone calls and multilingual speech.
Existing Speech-to-Text API customers get the upgrade without changing code.
Pricing stays $0.10 per hour of audio for batch work and $0.20 per hour for streaming, with speaker labels, timestamps, and key-term biasing included.
On one internal short-phrase test set, the company reports word error rate falling from 20.6 percent to 6.8 percent.
What is still unknown or disputed
- Independent audits beyond the cited Artificial Analysis leaderboard are not in the post.
Primary sources
Every claim in this story is drawn from the documents below. If a fetch failed, that is recorded on the card.
Source 1
Introducing Grok Voice Transcribe 2.0
xAI · September 18, 2026
Today we're releasing Grok Voice Transcribe 2.0, our latest speech-to-text model. … twice as accurate as Grok Voice Transcribe 1.0, at the same price. … On the public Artificial Analysis leaderboard, Grok Voice Transcribe 2.0 ranks first for accuracy among 32 streaming models. … Batch transcription remains $0.10 per hour of audio and streaming $0.20 per hour, with diarization, timestamps, and key terms included.
https://x.ai/news/grok-voice-transcribe-2