Google has expanded its conversational AI ecosystem with the launch of Gemini 3.8 Live with Live Avatar, adding near-real-time video output to its live audio models. The update enables virtual personas to listen, speak, and dynamically express themselves in synchronous conversations, targeting enterprise applications like automated customer care and guided digital walkthroughs.
The capability is now accessible directly via Gemini Enterprise.
Core Features and Capabilities of Gemini 3.8 Live with Live Avatar
Multimodal Real-Time Interaction
Traditional conversational AI interfaces rely on voice or text alone. Live Avatar combines speech-to-speech interaction with real-time video streaming, matching facial movements and lip-syncing directly to generated audio. The system processes visual and audio cues simultaneously to deliver natural turn-taking during dialogue.
Asynchronous Tool Execution
To maintain fluid conversation during backend operations, Gemini 3.8 Live uses asynchronous tool calling. The virtual avatar can query databases, run background tasks (such as hotel check-ins or booking confirmations), and fetch external information without pausing the voice interaction or interrupting the video stream.
Native Multilingual Synchronization
The platform supports real-time voice and video synchronization across 97 languages. It automatically adjusts facial expressions and lip movements to fit the phonetic nuances of each language mid-conversation, eliminating visual artifacts or sync degradation when switching dialects.
Custom Brand Avatars
Alongside a preset library of digital personas, enterprises can create branded avatars using a single reference image. This allows organizations to maintain visual consistency, character identity, or corporate branding. (Note: Custom avatar generation currently requires developer allowlisting via Google).
Security and Watermarking
To address identity risks and potential misattribution, Google integrates its SynthID watermarking tech directly into both the visual and audio streams. This imperceptible signature ensures generated media can be verified as synthetic content, helping organizations maintain transparency in customer-facing deployments.
Getting Started
Developers and organizations with access to Gemini Enterprise can deploy Gemini 3.8 Live with Live Avatar immediately using Google’s updated API documentation.