SpaceXAI Releases Grok Voice Transcribe 2.0

SpaceXAI released Grok Voice Transcribe 2.0, a speech-to-text model built on the audio foundation model behind Grok Voice. The new model ranks first for accuracy among 32 streaming models on the public Artificial Analysis leaderboard.

Across internal evaluations, word error rates improved compared to Grok Voice Transcribe 1.0. On short phrases such as in-car commands across 19 languages, the word error rate drops from 20.6% to 6.8%.

The model supports batch and streaming transcription, word-level timestamps, speaker diarization, multichannel transcription for up to 8 channels, key term biasing for up to 100 domain terms, text formatting, filler word removal, and smart turn detection.

Batch transcription costs $0.10 per hour of audio and streaming costs $0.20 per hour, matching version 1.0 pricing with diarization, timestamps, and key terms included. Grok Voice Transcribe 1.0 will be deprecated in the coming weeks.