Gemini 3.8 Live introduces Live Avatar for enhanced enterprise interactions

Google just gave its enterprise AI customers a face. Gemini 3.8 Live with Live Avatar, announced on September 24, 2026, introduces real-time video avatars that can lip-sync, express natural facial movements, and hold fluid conversations across 97 languages, all without the uncanny-valley lag that has historically made AI avatars feel like talking to a mannequin with a WiFi connection. The release builds on the Gemini 3.8 Live foundation launched just nine days earlier on September 15, which focused on low-latency voice processing and asynchronous workflows. Live Avatar takes those voice capabilities and wraps them in a visual layer designed specifically for enterprise use cases like interactive customer service and virtual product walkthroughs. What Live Avatar actually does At its core, Live Avatar is a real-time video generation system tightly coupled to Gemini’s speech-to-speech engine. When a user interacts with a Gemini-powered agent, they see a photorealistic avatar responding in real time, complete with synchronized lip movements, appropriate facial expressions, and natural turn-taking. The multilingual piece is particularly notable. Live Avatar supports 97 languages with sp...