Google launches Gemini 3.8 Live's Live Avatar real-time virtual avatar feature.

Beating AI News Flash: After releasing Gemini 3.8 Live last week, Google today launched Gemini 3.8 Live with Live Avatar, binding real-time voice conversation with low-latency streaming video generation, enabling AI agents in enterprise customer service, guided tours, and other scenarios to "listen, see, and speak," complete with lip sync, natural expressions, and smooth turn-taking. The feature is available in Gemini Enterprise starting today.

Live Avatar can process visuals and audio simultaneously, and when tools are called asynchronously in the background (such as hotel check-in), the conversation is not interrupted; it supports seamless switching between 97 languages, with lip sync and expressions changing accordingly. Enterprises can use preset avatars or customize brand avatars with reference images (customization currently requires an enterprise whitelist). All audio and video output carries a SynthID watermark, making it easy to identify as AI-generated content.