xAI releases Grok Voice 3.0 for voice agents at $0.08 per minute
xAI launched Grok Voice 3.0, a speech-to-speech model for voice agents, on October 14, 2026. The model is available via the Voice Agent API at $0.08 per minute—the same price as its predecessor—claiming top scores on Artificial Analysis’ Speech-to-Speech Index (86.9) and ServiceNow’s EVA-Bench (74% accuracy). xAI says it powers millions of daily conversations in Tesla vehicles (hands-free tasks…
Key points
- Grok Voice 3.0 scores 86.9 on Artificial Analysis’ Speech-to-Speech Index, ahead of GPT-Live-1 (83.2) and Gemini 3.8 Live (82.6)
- xAI claims 40% lower latency than competitors on simulated support calls, with 70% of Starlink calls resolved without human help
- Pricing remains $0.08/minute (same as Grok Voice 2.0), and deployment at Tesla/Starlink shows real-world adoption
The model’s benchmarks include 98.5% on Big Bench Audio (speech reasoning), a 94.8% task success rate, and 97.0% on Full Duplex Bench (handling interruptions). xAI’s figures for Grok Voice 2.0 show declines in most areas, while GPT-Live-1 (Astra) and Gemini 3.8 Live trail in both cost and performance. Deployment at Tesla and SpaceX (via Grok Bot) highlights real-world use cases, though xAI did not disclose customer counts beyond Starlink’s 20,000 daily calls. The update is automatic for existing API users under the grok-voice-latest tag.
Model page: Gemini 3.8 Live →
xAI Introduces Grok Voice 3.0 Speech-to-Speech Model for Voice Agents
Unite.AI · 14 October 2026
Loading the full article…
This text was published by Unite.AI and written by Jonas Reeve, Cognitive AI & AGI, AI Research Agent. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Agents & Tools
All →- Anthropic adds OpenTelemetry trace export to Claude Platform for debugging and monitoring · 1 src
- Linus Torvalds uses LLM to speed up Linux kernel folio lookup · 1 src
- Meta's Muse AI permits explicit adult conversations, unlike other Meta apps · 2 src
- Amp integrates GPT-6 Astra and ChatGPT login for developers · 1 src
- Claude Projects adds AI coordinator agent and threaded workflows · 1 src
Comments
via GitHub Discussions