OpenAI WebRTC Playground

Mis à jour le 2025-01-03Confiance : high

openai-webrtc-playgroundsimon-willisonbrowser-toolwebrtc-apigpt-realtime-2document-contextaudio-conversationplayground-developmentapi-testingmodel-comparison

Browser-based experimental tool created by simon-willison for testing and exploring openai's webrtc-api capabilities. Demonstrates playground-development approach to accessing advanced AI features before they appear in consumer applications.

Development History

December 2024: Initial version created to test OpenAI's WebRTC API for realtime audio models
June 2026: Updated to support gpt-realtime-2 model and document context functionality

Features

Model Selection

Support for multiple OpenAI audio models
Access to gpt-realtime-2 with "GPT-5-class reasoning"
Model comparison capabilities

Document Context Integration

Upload and paste document content
document-context-audio conversations
Exploration of information through conversational interfaces

Browser Integration

Direct webrtc-api access without additional software
Real-time audio streaming
Low-latency voice interactions

Significance

The playground addresses a key gap where advanced models like gpt-realtime-2 are available through API but not yet integrated into consumer applications like the ChatGPT iPhone app. This demonstrates the value of independent tooling for early access to cutting-edge capabilities.

Technical Implementation

Browser-based interface using webrtc-api
Real-time audio processing
Document content integration for conversational exploration
Model selection and configuration options