Which models are recommended for the Gemini Live API?

The recommended models are gemini-3.1-flash-live-preview for general real-time dialogue and gemini-3.5-live-translate-preview for real-time streaming translation.

Is function calling supported in the Live API?

Yes, but it is currently synchronous. The model will wait for the tool response before continuing its turn, allowing for real-time tool integration within a live session.

What is the best way to handle authentication in client-side apps?

For client-side deployments, always use ephemeral tokens to secure your application and never expose your primary API keys in the browser.

Does the Live API support WebRTC?

The native Gemini Live API currently only supports WebSockets. However, this skill provides information on partner integrations like LiveKit and Pipecat that facilitate WebRTC support.

How should I handle user interruptions in audio streams?

You should monitor for 'interrupted' signals in the server response content and immediately clear your local audio playback queues to ensure a natural conversational flow.

Gemini Live API Dev - Claude Code Skill for Real-time AI

This skill provides specialized guidance for building low-latency, multimodal applications with Gemini Live. It empowers developers to implement real-time audio and video streaming, voice activity detection (VAD), and sophisticated session management using the latest Python and JavaScript SDKs. Whether you are building an AI-powered voice assistant, a real-time translator, or an interactive video agent, this skill ensures you follow best practices for function calling, secure authentication with ephemeral tokens, and efficient context management.

Key Features

01Session management best practices including resumption and context compression.

02Synchronous function calling and Google Search grounding integration.

033,794 GitHub stars

04Support for Gemini 3.1 Flash Live and Gemini 3.5 Live Translate models.

05Implementation patterns for bidirectional audio and video streaming over WebSockets.

06Configuration for Voice Activity Detection (VAD) and interruption handling.

Use Cases

01Building natural AI voice assistants with low-latency mic-to-speaker interaction.

02Developing real-time streaming translation tools supporting over 70 languages.

03Creating vision-aware AI agents that process live camera or screen frames.

Tieline — product intent grounded in code

Gemini Live API Development

Key Features

Use Cases

Kotlin Ktor Server Patterns

Pr Preflight Production Risk Security Review

Python Design Patterns & Best Practices

Key Features

Use Cases