Introducing Gemini 3.8 Live with Live Avatar

Introducing Gemini 3.8 Live with Live Avatar

Gemini 3.8 Live expands the conversational AI platform with a Live Avatar feature that pairs near‑real‑time video generation with speech. The result is a dynamic visual persona that can listen, see, and speak, delivering a more natural, multimodal experience for enterprise applications such as customer support, virtual tours, and interactive training.

---

A richer, multimodal conversation

Human communication blends voice, facial expression, and body language. Live Avatar mirrors that blend by processing visual and audio inputs simultaneously and responding with synchronized video and audio output. The avatar’s lip‑sync, eye contact, and expressive gestures adapt in real time, making interactions feel less like a chatbot and more like a live human counterpart.

---

Continuous tool execution with visual presence

Beyond visual flair, the feature leverages Gemini’s advanced reasoning and asynchronous tool calling. While the avatar maintains a fluid dialogue, it can trigger background API calls, fetch data, or perform transactions without breaking the conversational flow. This enables complex workflows—such as checking a hotel guest in or retrieving account details—while the avatar remains visibly engaged.

---

Global, multilingual reach

Live Avatar supports 97 languages with native speech‑to‑speech synchronization. The system dynamically adjusts lip‑sync and facial expressions as the language switches, preserving video fidelity and avoiding visual drift. Enterprises can therefore deploy a single avatar across diverse markets without sacrificing the quality of the visual experience.

---

Customizable brand avatars

Brands often need a visual identity that aligns with their style guidelines. Gemini offers a library of preset avatars and a custom avatar creation workflow. By supplying a high‑resolution reference image, developers can generate a fully animated avatar that retains the original likeness and brand styling. Custom avatars are currently available through enterprise allow‑listing.

---

Trust, transparency, and safety

All video and audio output from Live Avatar carries an imperceptible SynthID watermark embedded directly into the media. This watermark makes AI‑generated content detectable, helping to mitigate misinformation and misattribution. The feature is built on Google’s broader safety framework, with detailed model cards and responsible‑deployment guidelines publicly available.

---

Getting started

Gemini 3.8 Live with Live Avatar is now available in Gemini Enterprise. Developers can explore the API documentation, experiment with preset avatars, or request custom avatar allow‑listing. The integration follows the same authentication and usage patterns as existing Gemini APIs, simplifying adoption for teams already using Gemini 3.8 Live.

---

Bottom line: Live Avatar brings visual presence to conversational AI, delivering multilingual, brand‑consistent, and safety‑aware interactions that feel genuinely human. Enterprises looking to elevate their digital engagements now have a turnkey solution that combines real‑time video, continuous tool execution, and robust safeguards—all within the familiar Gemini ecosystem.