Gemini 3.8 Live with Live Avatar gives Google’s AI a face

Google has introduced Live Avatar as part of its Gemini 3.8 Live update, allowing users to carry on conversations with the model while watching an animated persona respond in real time. The company says the avatar animates facial expressions and lip-syncs to dialogue as the model speaks. The Live Avatar feature supports 97 languages, and Google claims it can switch between them without compromising video quality or causing visual drift. In a demonstration shared by the company, an avatar is shown speaking in English and Japanese with mouth movements aligned to the spoken content, and the interface also displays information on-screen during the exchange. Google will provide a library of preset avatars for customers to select, and it also plans to let organizations create custom avatars for their own use. The Live Avatar output will include an invisible SynthID watermark, and Google says it is deploying safeguards intended to “respect identity.” The Live Avatar rollout follows Google’s reveal of the Gemini 3.8 Live model roughly a week earlier; the company has described that model as capable of processing visual inputs in near real-time. For now, Live Avatar is limited to enterprise customers on the Gemini platform.