OpenAI Realtime Agents logo

OpenAI Realtime Agents

Paid

Advanced agentic patterns for voice agents using OpenAI Realtime API

4.7
Inputs: audio, textOutputs: audio, text
Type
Saas
Company
OpenAI

About OpenAI Realtime Agents

OpenAI Realtime Agents is an open-source demonstration project that showcases advanced, agentic patterns for building voice agents using the OpenAI Realtime API and the OpenAI Agents SDK. Built as a Next.js TypeScript application, it implements two primary patterns: a Chat-Supervisor pattern where a realtime chat agent handles basic interactions and defers complex tasks to a more intelligent text-based supervisor model (e.g., gpt-4.1), and a Sequential Handoff pattern where specialized realtime agents transfer users between each other to handle specific intents, ideal for customer service scenarios. The project provides a low-latency, streaming voice interaction framework and includes setup instructions for running locally with an OpenAI API key.

Key Features

Demonstrates two agentic patterns: Chat-Supervisor and Sequential Handoff
Uses OpenAI Realtime API for low-latency, streaming voice interactions
Built on OpenAI Agents SDK for agent orchestration, state management, and event handling
Chat-Supervisor pattern: realtime agent handles basic tasks, supervisor model handles tool calls and complex responses
Sequential Handoff pattern: specialized realtime agents transfer users to handle specific intents
Next.js TypeScript app with easy local setup

Pros & Cons

Pros
  • Open source and free to use (only requires OpenAI API key for API costs)
  • Demonstrates production-relevant patterns like supervisor delegation and agent handoffs
  • Integrates seamlessly with OpenAI's Agents SDK for scalable agent management
  • Low-latency, real-time voice interaction via WebSockets
  • Easy to set up and run locally with provided instructions
Cons
  • Not a production-ready solution; intended as a demonstration and learning resource
  • Requires an OpenAI API key and incurs usage costs for Realtime API and language models
  • Documentation limited to the README; no extensive support channels

Best For

Building voice-based conversational agents with advanced orchestrationCustomer service systems that route users to specialized agent domainsPrototyping multi-agent voice interactions using OpenAI's realtime capabilitiesLearning and experimenting with agentic patterns for voice AI

Alternatives to OpenAI Realtime Agents

FAQ

What is OpenAI Realtime Agents?
It is an open-source demo project by OpenAI that shows advanced agentic patterns for voice agents using the Realtime API and the OpenAI Agents SDK.
What patterns are demonstrated?
Two patterns: Chat-Supervisor (a realtime agent with a text-based supervisor for tool calls) and Sequential Handoff (specialized agents that hand off users based on intent).
How do I set up and run the demo?
It is a Next.js TypeScript app. Install dependencies with 'npm i', add your OPENAI_API_KEY to a .env file, and start the server with 'npm run dev'. Open http://localhost:3000 in a browser.
Do I need to pay for anything?
The code is free and open source, but you need an OpenAI API key, and usage of the Realtime API and models incurs standard OpenAI API costs.