Grok Voice 3.0
Grok Voice 3.0 is xAI's speech-to-speech model for real-time voice agents, released on October 14, 2026. It replaces Grok Voice 2.0 (now deprecated) in the Voice Agent API at the same $0.08 per minute, and it also runs Grok voice…
Best for
Real-time voice agents for customer support and phone calls where low latency and task success matter.
Overview
Grok Voice 3.0 is xAI's speech-to-speech model for real-time voice agents, released on October 14, 2026. It replaces Grok Voice 2.0 (now deprecated) in the Voice Agent API at the same $0.08 per minute, and it also runs Grok voice in Tesla cars and Starlink's support lines.
Strengths and limits
- Top score on the Artificial Analysis Speech-to-Speech Index in xAI's launch table (86.9 vs 81.3 for Grok Voice 2.0)
- Sub-second latency per xAI, at the same per-minute price as Grok Voice 2.0
- The grok-voice-latest alias upgraded existing agents automatically, with no code changes
- Trails GPT-Live-1 (Astra) on the τ-Voice support-call benchmark (71.6% vs 74.5%) in xAI's own table
- Languages, voices and context window are not listed in xAI's docs; all benchmarks are vendor-reported
Published benchmark scores
Scores are as published by the source; settings (reasoning effort, tools, attempts) differ between labs, so compare with care. How to read benchmarks →