Google just released the Gemini 3.8 Live update. This new software lets users talk to the AI model while watching an animated persona respond in real time.
The feature is called Live Avatar. It lip-syncs and displays various facial expressions during active conversations.
Live Avatar is currently restricted to Gemini Enterprise customers. Google states the avatar handles 97 supported languages without degrading video fidelity or introducing visual drift.
A demonstration video published by Google shows the Live Avatar speaking in both English and Japanese. The mouth animations match the spoken words across both languages.
The system can also display relevant information on screen while talking.
This launch follows the introduction of the Gemini 3.8 Live model roughly one week prior. Google reports that the underlying model processes visual inputs in near real time.
Google provides a library of preset avatars for corporate customers. Organizations also have the option to build custom avatars.
Google applies its invisible SynthID watermark to all Live Avatar outputs. The company includes safeguards designed to respect identity.
