Gemini 3.8 Live debuts animated avatar supporting 97 languages

Google has integrated a new feature called Live Avatar into its Gemini 3.8 Live assistant, granting the AI an animated face capable of speaking, interacting, and displaying facial expressions. This update aims to make artificial intelligence conversations resemble video calls with real people.

Realistic Interaction Features

The avatar synchronizes lip movements with speech generated by Gemini while utilizing facial expressions to respond more smoothly during dialogue. This functionality reduces the perception that users are engaging with a system limited to text or audio inputs.

Live Avatar supports approximately 97 languages, automatically adjusting speech, lip movements, and expressions as the conversation progresses. Google provides a set of ready-made avatars and allows certain enterprise clients to create custom versions based on reference images, preserving the character’s features, identity, and visual style.

Enterprise Access and Background Tasks

The feature extends beyond realism, enabling the system to understand and interact with both audio and visual information simultaneously. This capability opens potential applications within companies, such as customer service, interactive tours, and experiences requiring highly responsive interfaces.

A key advantage is that the avatar does not necessarily pause when Gemini executes other tasks. The assistant can utilize tools, search for information, or run connected services in the background while maintaining the ongoing conversation.

Gemini 3.8 Live with Live Avatar is currently available through Gemini Enterprise, with custom avatar creation restricted to institutions granted access to this specific capability. Google confirmed that audio and video outputs from the system carry the SynthID watermark, marking content generated by artificial intelligence.