xAI / SpaceX Super IntelligenceGrok Voice
Grok Voice 3.0
Grok's real-time voice API exchanges speech over WebSockets and supports turn detection and tool use.
What the release provides
Client applications can connect with short-lived tokens instead of receiving a permanent API key. Session configuration controls the voice, instructions and turn detection.Grok Voice API overview
How to use it well
Practical guidance. Check interruption handling, silence and tool-call delays in a complete conversation. A voice answer must stop when the user interrupts, and an external action needs a clear confirmation. Network and playback delays are part of the experience.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | API |
| Model identifiers | grok-voice-3.0 |
| API model | grok-voice-3.0; real-time speech-to-speechGrok Voice 3.0 current model card |