Grok Voice Models by xAI
Last refreshed 2026-10-17. Next refresh: weekly.
Capabilities
Links
WebsiteAbout
xAI's Grok Voice family of full-duplex voice agent models for enterprise workflows. Designed for complex, multi-step customer support, sales, and enterprise applications with real-time reasoning, structured data capture, and broad language support. The family is evaluated on the τ-voice Bench leaderboard for full-duplex voice agents under realistic conditions.
Decision facts
- Best fit
- realtime voicespeech recognitionaudio
- Capability starting point
- Grok Voice Think Fast 1.0 with reasoning, JSON / Tool use, structured outputs, and multimodal inputs
- Lowest tracked input
- Not tracked
- Closest related family
- Grok 1
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
Use when the workload needs realtime voice, multimodal inputs, and audio.
Use when the workload needs speech recognition and audio.
Use when the workload needs realtime voice, reasoning, and JSON / Tool use.
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| Grok Voice 3.0 | Use when the workload needs realtime voice, multimodal inputs, and audio. | 2026-10 | realtime voicemultimodal inputsaudio | Current |
| Grok Voice Transcribe 2.0 | Use when the workload needs speech recognition and audio. | 2026-09 | speech recognitionaudio | Current |
| Grok Voice Think Fast 1.0 | Use when the workload needs realtime voice, reasoning, and JSON / Tool use. | 2026-04 | realtime voicereasoningJSON / Tool use | Current |
Release Timeline
4 release groupsReplaced By
Keep for legacy integrations; evaluate Grok Voice 3.0 before new work.
Specifications(4 models)
| Model | Released | Multimodal | Reasoning | JSON / Tool use | Structured Outputs |
|---|---|---|---|---|---|
| Grok Voice 3.0 | 2026-10 | Yes | No | No | No |
| Grok Voice Transcribe 2.0 | 2026-09 | No | No | No | No |
| Grok Voice Think Fast 1.0 | 2026-04 | Yes | Yes | Yes | Yes |
