Google launches Live Avatar for Gemini 3.8 Live
The feature pairs a real-time video persona with Gemini's dialogue models and is now generally available to Gemini Enterprise customers, though one outlet called the results "creepy."
Google said this week that Live Avatar, an animated video persona for its Gemini 3.8 Live and Live Extended Thinking models, is now generally available to Gemini Enterprise customers. The feature pairs low-latency video with the models' existing voice dialogue.
The avatars offer "precise lip-syncing, natural expressions and fluid turn-taking," Google said. Speech is synchronized across 97 languages, with adaptive lip movement. Customers can choose from a preset library or build branded characters from reference images. Every avatar's audio and video carries Google's SynthID watermark to flag it as AI-generated, TestingCatalog reported.
Google demonstrated the feature handling customer service and guided walkthroughs. In one demo, a hotel check-in scenario, the avatar keeps talking while background tool calls retrieve data. Another demo showed a conversation that switches languages mid-stream. Access currently requires enterprise allowlisting for custom avatars, TestingCatalog said.
Engadget, which covered the launch separately, called the avatars "creepy." The outlet said that given "people's distaste for AI slop," the feature may not be popular if it reaches consumer-facing products. Engadget said that is possible, even though today's rollout is enterprise-only.