Google has introduced Live Avatar for its Gemini 3.8 Live model, designed to provide enterprise customers with real-time video avatars. The system features lip-syncing, natural facial expressions, and support for 97 languages, allowing for fluid conversations without visual drift during language switching.
The new feature builds upon the Gemini 3.8 Live foundation, which was released on September 15, 2026, and focused on low-latency voice processing. Live Avatar integrates visual generation directly with the speech-to-speech engine to facilitate use cases such as customer service and product demonstrations.
To address security and compliance, Google implemented a verification process for custom avatars and integrated SynthID watermarking into all generated content. The platform also includes data governance controls tailored for industries like finance, healthcare, and government to ensure adherence to regulatory frameworks.
This release positions Google against competitors such as Microsoft and Meta, as well as specialized startups like Synthesia and HeyGen. By combining conversational intelligence with visual avatars in a single platform, Google aims to reduce the latency and complexity associated with using separate tools for scripts and AI models.
