Gemini 3.8 Live with Live Avatar gives Google’s AI a face

AI-generated avatars now speak and react in real-time across multiple languages, with customizable options becoming available.

Google has launched Gemini 3.8 Live with a new feature called Live Avatar, which provides an animated AI persona that converses with users in real time. This avatar can lip-sync and display facial expressions, supporting 97 languages with no degradation in video quality or visual drift. The feature, revealed shortly after the Gemini 3.8 Live model update that processes visual inputs rapidly, is currently exclusive to Gemini Enterprise customers. Users can select from preset avatars or create custom ones. Google states that the output includes an invisible SynthID watermark and safeguards for identity protection.

The integration of real-time animated avatars with advanced language models marks a significant step in human-AI interaction, moving beyond text-based interfaces to more engaging and potentially naturalistic communication. This could accelerate adoption in customer service, education, and virtual assistance, where a more human-like presence might foster greater trust and usability. The ability to support 97 languages seamlessly also signals a push towards global accessibility for these advanced AI tools.

While currently limited to enterprise clients, the availability of custom avatar creation suggests a future where businesses can brand their AI interactions distinctively. The inclusion of SynthID watermarking and identity safeguards indicates a proactive approach to addressing concerns around AI-generated content authenticity and misuse. This move sets a benchmark for the realism and responsiveness expected from conversational AI interfaces in the coming years.