Google Introduces Gemini 3.8 Live with Live Avatar for Real-Time Visual Conversations

Google introduced Gemini 3.8 Live with Live Avatar on September 24, 2026, combining Gemini’s live dialogue capabilities with low-latency streaming video for near-real-time visual interaction. Google says the feature builds on its Gemini 3.8 Live launch the previous week, enabling an avatar to listen, see, and respond with speech and a dynamic visual presence.

Live Avatar processes visual and audio inputs together and is designed with precise lip-syncing, natural expressions, and fluid turn-taking. It can also call tools and retrieve data in the background while dialogue continues. Google says speech synchronization works across 97 languages, with lip-sync and expressions adapting as conversations switch languages.

The feature is available in Gemini Enterprise. Organizations can use preset avatars or generate custom animated avatars from reference images, though custom avatar creation is currently limited to enterprise allowlisting. Google gives customer service and interactive walkthroughs as examples of possible enterprise uses.

Google says audio and video generated by its AI products are watermarked with SynthID to help identify AI-generated content. The launch brings real-time visual interaction, background tool use, and multilingual speech synchronization into an enterprise offering; Google’s announcement lists Gemini Enterprise as the availability channel.