Can companies use their own face or mascot as a Gemini Live Avatar?

Gemini 3.8 Live with Live Avatar ships with a library of preset avatars, and organizations can also build a custom one from a single high-quality reference image that keeps the likeness, brand style or character identity. Google limits custom avatar creation to enterprise allowlisting, with a strict verification process, so most teams start with the presets.

Where is Gemini 3.8 Live with Live Avatar available?

Google Cloud made Gemini 3.8 Live with Live Avatar generally available on September 24, 2026 in Gemini Enterprise, with US and EU endpoints. Developers try it in the multimodal Live section of Agent Platform Studio in the Cloud console and build on the Live API. Gemini 3.8 Live Extended Thinking stays in private preview on Gemini Enterprise.

Is video from Gemini Live Avatar marked as AI-generated?

Google embeds a SynthID watermark in all audio and video that Gemini 3.8 Live with Live Avatar generates. SynthID is an imperceptible signal that helps tools detect AI-generated content later, so a recording of an avatar agent can be identified as machine-made even when the viewer cannot tell by eye.

Who is already using Gemini 3.8 Live on Google Cloud?

Google Cloud names three early users of Gemini 3.8 Live. Cox Automotive built an Autotrader shopping assistant that highlights parts of the screen as it talks. Equal AI handles more than 1 million live calls a day across 9 Indian languages. Salesforce connects the model to Agentforce for service agents.

Google DeepMind · 2026-09-24 · major

Gemini 3.8 Live with Live Avatar — Google's voice agents get a talking face

Gemini 3.8 Live with Live Avatar adds a real-time video persona to Google's voice model, with lip-sync in 97 languages. It is generally available in Gemini Enterprise, and every frame carries a SynthID watermark.

Gemini 3.8 Live can now appear on screen as a lip-synced video avatar that talks, listens and uses tools in real time.

Quick facts

Maker	Google DeepMind / Google Cloud
Status	Generally available (Sept 24, 2026)
Where	Gemini Enterprise, Agent Platform Studio
Regions	US and EU endpoints
Languages	97, with lip-sync
Custom avatars	Enterprise allowlist only
Watermark	SynthID on all audio and video

What is it?

Live Avatar is a new option for Gemini 3.8 Live, Google's speech-to-speech model: the agent now shows up as an animated video persona instead of a voice alone. Google Cloud made it generally available on September 24, 2026 in Gemini Enterprise. Teams pick a preset avatar or, if allowlisted, build a custom one from one reference image.

How does it work?

Google couples the model's live dialogue with low-latency streaming video generation, so lip movements and facial expressions are produced in step with the speech. The multilingual speech-to-speech sync covers 97 languages without visual drift. Tool calls run in the background, so the avatar keeps talking while it fetches data, and the model can still read camera feeds and screen shares.

Why does it matter?

Customer-service and walkthrough agents usually need a separate avatar vendor stitched onto a voice model, which adds delay and cost. With a face built into the same Gemini session, a support bot or product guide can look and sound like one system, in one API call path, with watermarking already applied.

Who is it for?

teams building customer-facing voice agents

Frequently asked questions

Can companies use their own face or mascot as a Gemini Live Avatar?: Gemini 3.8 Live with Live Avatar ships with a library of preset avatars, and organizations can also build a custom one from a single high-quality reference image that keeps the likeness, brand style or character identity. Google limits custom avatar creation to enterprise allowlisting, with a strict verification process, so most teams start with the presets.
Where is Gemini 3.8 Live with Live Avatar available?: Google Cloud made Gemini 3.8 Live with Live Avatar generally available on September 24, 2026 in Gemini Enterprise, with US and EU endpoints. Developers try it in the multimodal Live section of Agent Platform Studio in the Cloud console and build on the Live API. Gemini 3.8 Live Extended Thinking stays in private preview on Gemini Enterprise.
Is video from Gemini Live Avatar marked as AI-generated?: Google embeds a SynthID watermark in all audio and video that Gemini 3.8 Live with Live Avatar generates. SynthID is an imperceptible signal that helps tools detect AI-generated content later, so a recording of an avatar agent can be identified as machine-made even when the viewer cannot tell by eye.
Who is already using Gemini 3.8 Live on Google Cloud?: Google Cloud names three early users of Gemini 3.8 Live. Cox Automotive built an Autotrader shopping assistant that highlights parts of the screen as it talks. Equal AI handles more than 1 million live calls a day across 9 Indian languages. Salesforce connects the model to Agentforce for service agents.

Try it

https://console.cloud.google.com/agent-platform/studio/multimodal-live