Google Makes Gemini 3.8 Live with Live Avatar Available to Enterprise Customers
Google announced the general availability of Gemini 3.8 Live with Live Avatar in Gemini Enterprise. The service adds talking video avatars with synchronized synthetic speech to conversational agents.
On September 24, 2026, Google announced the general availability of Gemini Live Avatar as part of Gemini Enterprise. The Gemini 3.8 Live with Live Avatar feature allows enterprise developers to add a talking visual character to voice and video agents without the need for their own rendering system.
The avatar creates video of a talking character synchronized with synthetic speech. Google says it supports prebuilt avatars, automatic switching between 97 languages, and asynchronous tool calling during conversations.
What Gemini Live Avatar brings
The new feature is designed for Gemini Enterprise and expands the conversational AI interface with a video layer. An agent therefore does not have to communicate only through voice or text, but can respond through an animated character with spoken delivery.
According to Google, the system includes synchronization between the avatar’s voice and video. The platform supports prebuilt characters that developers can use when creating their own agents. Asynchronous tool calling is intended to allow the agent to work with tools while conducting a conversation.
Google says the service supports automatic switching between 97 languages. However, independent comparative results for the quality of language switching, lip synchronization, or responsiveness in real-world enterprise deployments remain undisclosed.
Custom avatars are restricted
Creating a custom avatar from a reference image is not available to all Gemini Enterprise customers. Google requires an enterprise allowlist and verification for this option. The restriction applies specifically when the system is expected to work with a particular reference photo instead of a prebuilt avatar.
Both generated audio and video contain invisible SynthID watermarks. Google has thus included a mechanism for labeling synthetic content in the service. The fact that custom avatars require additional access and verification also highlights the importance of risks associated with identity misuse in video agents.
A production tool for enterprise AI agents
General availability means that Gemini Live Avatar is a production-ready tool for enterprise developers in the Gemini Enterprise ecosystem. They can use it with agents that combine synthetic speech, a visual character, and calls to additional tools.
Google describes the interaction as natural and low-latency and mentions potential uses such as customer support. However, these are product claims by the company, not independently verified results from comparative tests.
Google has not yet confirmed whether or when Live Avatar will also become available to regular users outside Gemini Enterprise. Going forward, it will be important to monitor availability by region, pricing and API limits, expanded access beyond the enterprise product, and independent tests of latency, synchronization, and safeguards for custom avatars.
Sources
- Google Cloud Blog – Google confirms general availability in Gemini Enterprise, enterprise deployment, endpoints in the US and EU, and the restriction of custom avatars to an allowlist.
- Google Blog – Google describes the video avatar, speech and facial-expression synchronization, 97 languages, asynchronous tool calling, and SynthID watermarks.
- Google Cloud Documentation – The documentation confirms Live Avatar configuration through the Gemini Live API and rules for using custom reference images.
- The Verge – Independently summarizes that Live Avatar provides an animated character with lip synchronization and expressions and is intended for Gemini Enterprise customers.
Verified and updated: 09/25/2026 06:21
