OpenAI API Playground
A native SwiftUI playground for testing models, streaming events, and running MCP tools.
OpenResponses is a native SwiftUI playground for testing OpenAI models, streaming responses, and managing tool executions (including secure sandboxed Python execution, browser automation, file searches, and custom MCP server endpoints).
Core Features
Multi-Model Support
Access GPT-5, GPT-4.1, O-series reasoning models, and specialized computer-use models with dynamic selection and compatibility checking.
Enhanced Streaming
Real-time responses with granular status updates showing AI thinking, web searching, code generation, and tool execution progress.
Computer Use
Production-ready browser automation allowing the AI to navigate, click, type, and interact with web content with safety approvals.
Web Search
Access up-to-date information from the internet with both standard and preview search capabilities integrated seamlessly.
Code Interpreter
Execute Python code in secure sandboxed environments (auto, secure, gpu) with file preloading and rich artifact display.
Image Generation
Create images from text prompts using gpt-image-1 with real-time streaming previews and configurable quality settings.
File Search
Search across multiple vector stores simultaneously with support for 43+ file types including documents, code, and archives.
MCP Integration
Connect to Model Context Protocol servers (GitHub, Notion, Slack, etc.) with secure authentication and approval workflows.
Advanced Controls
Fine-tune temperature, tokens, penalties, reasoning effort, and other parameters with full API transparency.
Developer Tools
API inspector, debug console, analytics dashboard, and comprehensive logging for troubleshooting and monitoring.
Prompt Library
Save and manage reusable prompt configurations with all settings for quick switching between different tasks.
Conversation Management
Local conversation storage with export capabilities, multi-conversation support, and comprehensive message management.
Advanced Capabilities
Real-Time Monitoring
Track streaming events, tool executions, reasoning phases, and token usage with expandable activity feeds and detailed status indicators.
Security & Safety
User approval workflows for MCP tool calls and computer actions, secure keychain credential storage, and comprehensive safety checks.
Native Experience
Built with SwiftUI for iOS, iPadOS, macOS, and visionOS with full accessibility support, dynamic type, and keyboard navigation.
Rich Content
Embedded web content, enhanced image display, Markdown rendering with syntax highlighting, and support for 43+ file types.
Architecture Highlights
MVVM Pattern
Feature-oriented structure with clear separation of concerns, comprehensive service layer, and dependency injection.
Streaming Events
Handles 40+ streaming event types from OpenAI's Responses API with dedicated parsing and real-time UI updates.
Data Management
Local conversation persistence, secure keychain storage, vector store management, and comprehensive file handling.