Google's Gemini 3.8 Live adds an animated avatar that lip-syncs in 97 languages

Google has given its Gemini assistant a visible face. As reported by The Verge, the company's Gemini 3.8 Live update now includes "Live Avatar," an animated on-screen persona that holds conversations with users in real time, lip-syncing its speech and shifting its facial expressions as it responds.

The rollout is narrow for now. Live Avatar is available only to Gemini Enterprise customers, meaning Google's business-facing tier gets the feature before the consumer Gemini app does.

How the avatar behaves

According to Google, the avatar talks with users while an animated character reacts on screen, moving its mouth in time with what it says and changing expression as the conversation unfolds. The persona can also pull information up on screen while it speaks, layering visual answers on top of the spoken exchange.

Language switching appears to be a central part of the pitch. Google says Live Avatar can move between the 97 languages it supports "without degrading video fidelity or introducing visual drift." In a demonstration video shared by the company, the avatar speaks in both English and Japanese, and its mouth animation lines up with whichever language it is using at the moment.

Built on the new Gemini 3.8 Live model

The Verge notes that Live Avatar arrives roughly a week after Google introduced the Gemini 3.8 Live model itself, which the company says processes visual inputs in near real time. That underlying capability is what allows the avatar to sustain a live, responsive conversation instead of playing back pre-scripted animation.

Organizations will not be stuck with a single character design. Google says it will offer a library of preset avatars for customers to pick from, and it will also let organizations build avatars of their own.

Watermarks and identity safeguards

A talking, animated AI persona inevitably raises questions about synthetic media and impersonation, and Google says it has taken steps on both fronts. Output from Live Avatar carries the company's invisible SynthID watermark, and Google says the feature ships with safeguards intended to "respect identity."

Why it matters

Live Avatar marks a shift in how Google presents real-time AI conversation: rather than a disembodied voice or a text box, users get an embodied on-screen presence that emotes and reacts. That format can make long interactions feel more natural, but it also deepens the stakes around trust, since a face and natural expressions make synthetic output more persuasive.

The enterprise-first launch suggests Google is treating avatar-based conversation as a business product to prove out before any wider consumer release, and the custom-avatar option points toward branded assistants for companies. The inclusion of SynthID watermarking signals that Google anticipates concerns about where this generated video ends up — even if the company has said little so far about how its identity safeguards will work in practice. For now, Live Avatar is the most visible step yet by a major AI provider to make a chatbot something you watch as well as talk to.