Introducing Gemini 3.8 Live with Live Avatar
Google is adding low-latency video avatars to Gemini’s live enterprise agents.
Gemini 3.8 Live with Live Avatar pairs speech with near real-time generated video, including lip sync, expressions, and turn-taking. Google says it can process audio and visual inputs together, keep talking while background tool calls run, and switch across 97 languages without video drift. Enterprises get preset avatars, with custom avatars available through allowlisting. All generated audio and video output is watermarked with SynthID. Google DeepMind's note
Gemini 3.8 Live with Live Avatar pairs speech with near real-time generated video, including lip sync, expressions, and turn-taking. Google says it can process audio and visual inputs together, keep talking while background tool calls run, and switch across 97 languages without video drift. Enterprises get preset avatars, with custom avatars available through allowlisting. All generated audio and video output is watermarked with SynthID. Google DeepMind's note
score 8