Google Adds Animated Avatar to Gemini for Real-Time Conversational AI

Google has rolled out a new feature for its Gemini AI assistant: an animated avatar that engages with users through lifelike facial movements and synchronized speech. The addition, part of the Gemini 3.8 Live update, brings a more human-like interface to AI conversations by rendering expressions and mouth movements that correspond to what the model is saying.

How the Avatar Works#

The Live Avatar feature displays an animated face that responds dynamically during interactions. The system lip-syncs to the AI's spoken output and shifts between different facial expressions to create a more natural conversational experience. The avatar can also display information and graphics on-screen simultaneously while speaking.

Multilingual and Customizable#

One of the system's strengths is its ability to handle language transitions smoothly. Google's avatar maintains visual consistency across 97 supported languages without losing video quality or experiencing visual glitches when switching between them. Demonstrations show the avatar seamlessly transitioning between English and Japanese, with mouth movements staying accurate to the phonetics of each language.

While Google will provide a collection of pre-built avatars for users to select, the feature also allows organizations to create custom avatars tailored to their brand or needs. All avatar output includes an invisible SynthID watermark, Google's content authentication system, along with other protections designed to prevent misuse of synthetic identity representations.

Availability and Context#

This launch follows Google's introduction of the broader Gemini 3.8 Live model, which the company developed to handle visual information with minimal latency. At present, the Live Avatar capability is restricted to Gemini Enterprise customers and is not yet available to general users.