Grok Voice Transcribe 2.0 Cuts Short-Phrase Errors; Price Stays Put
xAI shipped Grok Voice Transcribe 2.0 at unchanged $0.10 batch / $0.20 streaming per hour. Lab short-phrase WER falls from 20.6% to 6.8%; xAI cites first place on Artificial Analysis’s streaming accuracy board among 32 models.
Correction / update, October 14, 2026: The launch post announced a future default-model change. The current default comes from documentation updated October 3, not a confirmed September 18 switch.
Transcription price did not move. The claim that matters is the error cut on short phrases — and whether Artificial Analysis’s streaming board still shows that rank once other labs answer.
xAI (September 18): Grok Voice Transcribe 2.0, a speech-to-text model built on the audio foundation model behind Grok Voice. xAI says that across its real-world evaluations it is one of the most accurate transcription models available, and twice as accurate as Grok Voice Transcribe 1.0, at the same price. On the company’s short-phrase set, word error rate drops from 20.6% to 6.8%.
Price and availability#
Batch $0.10 per hour of audio, streaming $0.20 per hour, with diarization, timestamps, and key terms included. The September 18 announcement said 2.0 would soon become the default, with 1.0 deprecated in the following weeks; users temporarily keeping 1.0 were told to pin it explicitly. The no-code-change claim applies to default-model integrations. Documentation updated October 3 lists grok-voice-transcribe-2.0 as the default, us-east-1, batch/streaming support and keyterm prompting.
Independent ranking#
xAI says that on the public Artificial Analysis leaderboard, Transcribe 2.0 ranks first for accuracy among 32 streaming models. That ranking is attributed to xAI’s account of AA; this article has not independently verified the live board. This is a transcription model, not a chat model, so text-arena Elo does not apply.