Google has launched Gemini 3.8 Live with a new "Live Avatar" feature that enables users to converse with its AI model whilst watching an animated persona respond in real time. The avatar lip-syncs and displays facial expressions during conversations, bringing a more natural, human-like interface to AI interaction. This development represents a significant step towards more engaging and immersive AI interfaces, though access is currently restricted to Gemini Enterprise customers.

The Live Avatar supports 97 languages and can seamlessly transition between them without losing video quality or visual consistency. Google offers a library of preset avatars for users to choose from, whilst also enabling organisations to create custom avatars tailored to their needs. The feature includes Google's invisible SynthID watermark and identity safeguards, with the avatar capable of displaying on-screen information whilst speaking to provide additional context during conversations.