Google makes Live Avatar generally available in Gemini Enterprise
The feature layers generated video, lip-sync and facial expressions onto Gemini 3.8 Live's speech model, and it supports 97 languages, Google said in a Cloud blog post.

Key takeaways
- Google said in a Cloud blog post dated September 24, 2026 that Live Avatar is now generally available in Gemini Enterprise, after first appearing in preview at Google Cloud Next 2026.
- Live Avatar understands and speaks 97 languages with automatic detection, adapting lip-sync and facial expressions when a conversation switches languages, according to Fonearena's report on the rollout.
- Every generated audio and video stream carries a SynthID watermark, and custom avatar creation is limited to enterprises approved through an allowlist and verification process, Google said.
Google's Gemini 3.8 Live with Live Avatar is now generally available in Gemini Enterprise, the company said in a Cloud blog post dated September 24, 2026.
The post says Live Avatar was first previewed at Google Cloud Next 2026 and now moves into enterprise production. It follows the launch of Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, which Google announced the week before; Extended Thinking itself remains in private preview.
Live Avatar layers real-time generated video, with synchronized lip-syncing and facial expressions, onto Gemini 3.8 Live's native speech-to-speech model. Google says the combination lets a video avatar listen, watch and speak during a conversation across web, mobile and interactive kiosk apps, while executing tool calls and API requests in the background so it can keep talking while a task finishes. According to Fonearena, the system understands and speaks 97 languages with automatic detection, and it adapts lip-sync and facial expressions on the fly when a conversation switches languages, without introducing visual drift. Live Avatar can also process live camera feeds and screen shares alongside audio at the same time, Google said.
For identity protection, Google is offering a library of curated, pre-built avatars; building a custom avatar is restricted to enterprises approved through an allowlisting and verification process, according to the blog post. Every generated audio and video stream carries an imperceptible SynthID watermark, which Google says is meant to keep AI-generated output detectable and traceable. The feature ships with US and EU endpoints, provisioned throughput, enterprise compliance tooling and data governance controls, though Google's post did not disclose a separate price for Live Avatar.
This is a conversational tool built for enterprise customer contact, not a video generator for creators. Google's own examples, per the blog, show a step-by-step custom-avatar builder and a claims-intake agent that uses live video to process a customer's documents, not a standalone clip a filmmaker could export and edit. Anyone hoping to use Live Avatar for creative projects will not find it in a consumer app; it lives inside Gemini Enterprise's Agent Platform Studio and its API.
Live Avatar is available now to Gemini Enterprise customers through the Agent Platform Studio and via API documentation; it is not offered as a standalone consumer or creator product.
Sources
- cloud.google.com - Google's own GA announcement, date, features, allowlist, SynthID, availability
- fonearena.com - 97-language dynamic lip-sync detail, availability link
- qazinform.com - wire report citing Google's blog, kiosk and mobile use detail
- deccanherald.com - corroborates SynthID watermarking and rollout detail