Grok Voice Agent API at a glance
Developers can build real-time voice agents on the stack behind Grok in Tesla vehicles, with tool calling, web and X search, dozens of languages, and compatibility with the OpenAI Realtime API format.
What changed with Grok Voice Agent API
On December 17, 2025, xAI launched the Grok Voice Agent API, opening the voice stack it uses in the Grok apps and in Tesla vehicles to all developers. xAI says it built the voice activity detection, tokenizer, and audio models in-house. Agents can call custom tools, use xAI's real-time web and X search, and answer in the caller's language, switching mid-conversation when needed.
Pricing at launch was a flat $0.05 per minute of audio. The API follows the OpenAI Realtime API specification and has an official LiveKit plugin, with voices including Ara, Eve, and Leo. At launch, xAI said the API ranked first on the Big Bench Audio reasoning test, independently verified by Artificial Analysis, with average time to first audio under one second.
Why Grok Voice Agent API matters for businesses
For agencies building phone receptionists or support lines, a Realtime-compatible API means existing voice code can often be pointed at Grok with few changes. Later Grok Voice models were listed at $0.08 per minute, so confirm current pricing and test on real telephony audio before switching client traffic.
Put it to work: AI Voice Agents
Sources for this story
- Official
- Release notes (opens in a new tab)Official
xAI Docs · docs.x.ai
Last verified: . IA Compa summarizes public announcements for business readers. We are not affiliated with the companies mentioned. Always confirm details, pricing, and availability with the official source.
- #voice-agents
- #xai
- #grok
- #realtime-api
- #pricing
