The Live API Web Console is a React-based starter application developed by Google that enables developers to interact with the Gemini Live API through websocket connections. Built with TypeScript, the project provides a foundation for building real-time, interactive applications that leverage Google's multimodal AI capabilities. The application is accessible at ai.google.dev/api/live and serves as both a working example and a development platform for exploring the Live API's features.
The core functionality centers on establishing websocket communication with the Gemini Live API and handling real-time interactions. The application includes several key modules that handle distinct aspects of the development experience. A streaming audio playback module manages audio output, while media recording capabilities allow users to capture input from microphones, webcams, and screen capture. The unified log view provides developers with visibility into the communication flow and helps with debugging during application development. The project is structured around an event-emitting websocket client that facilitates communication between the websocket layer and the React frontend, along with a communication layer for processing bidirectional audio.
The repository includes multiple example implementations across different branches, demonstrating various use cases. The demos/proactive-audio branch showcases the Live API's proactive audio feature using the GenAI SDK. Original demonstration applications include GenExplainer, GenWeather, and GenList, each illustrating different approaches to building with the Live API. An example in the documentation demonstrates integration with Google Search grounding and data visualization using vega-embed, showing how developers can combine the Live API with external services and visualization libraries.
Getting started requires creating a free Gemini API key from aistudio.google.com and configuring it in the project's environment file. The project was bootstrapped with Create React App and includes standard npm scripts for development and production builds. The npm start command launches the development server on localhost:3000 with hot reloading, while npm run build creates an optimized production bundle.
According to GitGenius activity tracking, the repository shows consistent engagement with a median issue and pull request response latency of 91.9 hours across 63 tracked items. The most frequently addressed issue categories are bugs with 9 occurrences, followed by model enhancements and general enhancements with 6 each. The primary contributor, hapticdata, has logged 110 events, with additional significant contributions from ViaAnthroposBenevolentia and OptionIA. The repository shares contributors with golang/go, google-gemini/gemini-cli, and vercel/ai, indicating cross-project collaboration within the broader developer ecosystem.
The project explicitly states it is an experiment showcasing the Live API rather than an official Google product, with the developers committing to best-effort support and maintenance. The repository encourages open-source learning while requesting respect for intellectual property rights when sharing or creating derivative works.
