SpaceXSI Launches Grok Voice 3.0 Speech Model at $0.08/Min With Sub-Second Latency

SpaceXSI announced Grok Voice 3.0 on October 14, 2026, making the speech-to-speech model available in its Voice Agent API at $0.08/min ($4.80/hr) — the same price as Grok Voice 2.0. Agents configured with grok-voice-latest were switched over automatically. According to Artificial Analysis testing cited by SpaceXSI, Grok Voice 3.0 ranks #1 on the Speech-to-Speech Index with a score of 86.9 and scores 97% on Full Duplex Bench for handling pauses, turn-taking, and interruptions. On the τ-Voice benchmark for noisy, multi-accent support calls, Grok Voice 3.0 scored 71.6% — trailing GPT-Live-1 (Astra, Medium) at 74.5%, while beating Gemini 3.8 Live at 68.6% and Grok Voice 2.0 at 56.5% — at 40% lower latency than GPT-Live-1 and a lower cost per task on Big Bench Audio than both GPT-Live-1 and GPT-Realtime-2.1. The model also recorded a 74% accuracy score on ServiceNow's EVA-Bench across more than 200 enterprise scenarios. In production, Grok Voice 3.0 handles millions of daily conversations across Tesla $TSLA vehicles and answers 20,000 daily Starlink call center inquiries across multiple languages, resolving roughly 70% without human intervention and converting 1 in 4 sales calls into a purchase before hang-up.