Artificial Intelligence
Google launches Gemini 3.8 Live with Live Avatar
Google introduced Gemini 3.8 Live with Live Avatar, saying it adds a visual presence to live dialogue, supports background tool use and multilingual switching, and is now available in Gemini Enterprise.
- Google introduced Gemini 3.8 Live with Live Avatar on 24 September 2026.
- Google said the feature combined near real-time video generation with speech to create a visual persona for live dialogue.
- Google said Live Avatar became available in Gemini Enterprise and could keep talking while tools ran in the background.
- Google said the system could move across 97 languages and that custom avatar creation was limited to enterprise allowlisting.
- Google said SynthID watermarked the audio and video from its AI products to help detect generated content.
Google introduced Gemini 3.8 Live with Live Avatar, building on last week’s Gemini 3.8 Live launch. It said the new feature added near real-time visual presence to its native live dialogue models.
The company said the system paired video generation with speech so that the avatar could listen, see and speak. It also said the feature was intended for enterprise uses such as customer service and interactive walkthroughs.
Google said Live Avatar was available in Gemini Enterprise from the day of publication. It said asynchronous tool calling let the system fetch data in the background while the conversation continued.
The company said the feature supported native speech-to-speech synchronisation across 97 languages. It also said developers could create custom avatars from reference images, with that option limited to enterprise allowlisting.
Google said its AI output was watermarked with SynthID, which it said would help make generated audio and video detectable. It said the approach was meant to support identity protection, transparency and responsible deployment.
Named in this story
Companies
- introduced the feature and described its safeguards
Products and systems
- Gemini 3.8 Live
- the live dialogue model the feature extended
- Live Avatar
- the feature that added a visual persona to live conversations
- Gemini Enterprise
- the product where the feature became available
- SynthID
- the watermarking system used on AI output
How the source tells it
The piece was written as a product launch, with upbeat feature claims and a brief safety note to counterbalance the promotion.
- vendor boosterism benefit-led launch copy and repeated claims about richer enterprise use
- reassurance a closing section stresses safeguards, watermarking and transparency