Gemini 3.8 Live Avatar: Google has launched its real-time AI model, Gemini 3.8 Live with Live Avatar. This feature gives the AI not just a voice, but also a visual face. The Live Avatar can listen to the user and interpret visual input via the camera, while responding through both voice and video in near real-time. According to Google, the facial expressions and lip movements are synchronized with the spoken words.
Google has introduced this feature specifically for enterprise and business users. The company states that Gemini 3.8 Live with Live Avatar is now available within Gemini Enterprise. It can be utilized for tasks such as customer support, virtual assistants, and interactive product or service walkthroughs.
AI will see, hear, and respond in real-time
A key feature of Live Avatar is its multimodal experience; it can process audio and visual inputs simultaneously. This means that during a conversation with a user, it can understand the camera feed or items displayed on the screen and respond within that context. According to Google, it can also execute tools and API calls in the background during the conversation, ensuring the dialogue remains uninterrupted while tasks are being completed.
For instance, if a user asks the AI to check information or complete a task, the model can continue the conversation while performing that task in the background.
Conversation in 97 languages and adaptive expressions
Gemini 3.8 Live with Live Avatar also includes multilingual support. According to Google, it can switch between 97 languages during a conversation. As the language changes, the avatar’s lip-sync and facial expressions adjust accordingly.
Companies will be able to create brand-specific AI avatars
Google has made several pre-built avatars available for Live Avatar. Additionally, companies can create custom avatars tailored to their brand or characters. According to Google, animated and responsive avatars can be generated based on high-quality reference images. However, the ability to create custom avatars is currently limited to select organizations via enterprise allowlisting.
SynthID Watermark for AI Video and Audio
Amid growing concerns regarding the use of AI-generated fake videos and audio, Google has incorporated the SynthID digital watermark into Live Avatar. According to the company, this watermark is embedded in AI-generated audio and video outputs. While imperceptible to the human eye and ear, it can assist in identifying AI-generated content.
Currently Available for Business Users
Google has currently made Gemini 3.8 Live with Live Avatar available within Gemini Enterprise. The primary focus is on enabling companies to create highly interactive AI agents and virtual services.
