Grok Voice Transcribe 1.0

The first-generation speech-to-text model, still offered alongside grok-voice-transcribe-2.0 and NOT deprecated: xAI's release note of 2026-09-17 presents the pair as a choice - 'Use grok-voice-transcribe-1.0 or grok-voice-transcribe-2.0' - and names no retirement date, no deprecation and no successor, so those fields stay null rather than being inferred from the arrival of a version 2. Billed by duration of audio, not by token, so the token price fields are null by design. SAME RATE AS 2.0 ON BOTH AXES, and that is first-party twice over: xAI's pricing page publishes one Speech to Text line for the mode rather than one per model ($0.10 per hour REST, $0.20 per hour streaming), and the models-index hydration payload carries byte-identical integers for the two models ('perAudioSecond': '277778', 'perAudioSecondStreaming': '555556') in the 1e-10 USD-per-second units this catalog calibrated against rows it already held. Rate limits at the base tier are also identical: 10 rps / 600 rpm / 100 concurrent sessions. RELEASED IS NULL AS A SOURCED ABSENCE: xAI's release notes date the Speech to Text API going generally available to 2026-04-15 ('The xAI Speech to Text API is now generally available. Transcribe audio to text in 25 languages with batch and streaming modes'), but that entry does not name this model slug and no entry on the page does, so the API has a first-party date and the model does not. Added to the catalog 2026-09-19 as a completeness fix; xAI published nothing on this date.

Source: xAI official documentation ↗

Explore more

Grok Voice Transcribe 1.0 — questions & answers