Gemini 3.8 Live now lets users hold conversations with Google’s AI while watching an animated persona respond in real time. The new Live Avatar feature lip-syncs and displays a range of facial expressions during exchanges, though it is currently available only to Gemini Enterprise customers.
According to Google, the Live Avatar can move between the 97 languages it supports without degrading video fidelity or introducing visual drift. One demonstration shows the avatar speaking in English and Japanese, with its mouth animation matching the words in both languages. The persona can also bring up information on-screen as it talks.
How Live Avatar works
The launch follows the reveal of the wider Gemini 3.8 Live model roughly a week earlier, which Google says processes visual inputs in near real-time. This capability underpins the responsive, face-to-face style of interaction that Live Avatar is designed to deliver.
Google will provide a library of preset avatars for customers to choose from, and it will also allow organisations to create their own personas. This gives enterprise users flexibility in how they present the AI to their staff or clients.
Safeguards and availability
Output produced by the Live Avatar carries Google’s invisible SynthID watermark, part of a set of safeguards intended to respect identity. The watermark is designed to help identify AI-generated content without being visible to viewers.
For now, access to the Live Avatar feature remains restricted to Gemini Enterprise customers, with support spanning 97 languages.
Source
Image: theverge.com
