v1.2.0: OpenAI Responses API Integration & Dual-Mode Support

🚀 Major Release: OpenAI Responses API Integration

✨ New Features

🆕 OpenAI Responses API Integration

• Full support for 2025 Responses API with gpt-4o + image_generation tool
• Native multimodal understanding with better context awareness
• Enhanced prompt following and superior text rendering in images
• Real-time streaming with partial previews via Responses API

🔄 Dual API Architecture

• Seamless API switching via API_MODE environment variable
• Default to Responses API (recommended) with fallback to Images API
• Backward compatibility with existing Images API implementations
• Unified interface supporting both legacy and modern workflows

🖼️ Enhanced Image Editing

• Multi-image input support with proper message formatting
• Context-aware editing with conversation history integration
• Previous response ID linking for multi-turn editing sessions
• Advanced mask support for precise inpainting operations

🧪 Comprehensive Test Coverage

• 99+ tests with full API coverage for both implementations
• Complete error handling and edge case coverage
• Real API validation with integration testing
• Mock-based testing with proper dependency injection

🔧 Technical Improvements

⚙️ Enhanced Configuration System

• API_MODE: Switch between 'responses' (default) and 'images'
• RESPONSES_MODEL: Model selection for Responses API (default: gpt-4o)
• Flexible model switching between dedicated and integrated approaches

🏗️ SOLID Architecture

• Complete refactoring following dependency injection patterns
• Single Responsibility principle with focused services
• Interface-based design for maximum extensibility
• Test-driven development with comprehensive coverage

📚 Updated Documentation

📖 Comprehensive Guides

• Dual API documentation with feature comparison table
• Enhanced environment variables documentation
• Updated streaming examples for both API modes
• API mode selection guidance and best practices

📋 API Comparison

Feature	Responses API (gpt-4o)	Images API (gpt-image-1)
Latest Technology	✅ 2025 Responses API	⚠️ Legacy API
Text in Images	✅ Superior	✅ Good
Context Awareness	✅ Excellent	⚠️ Limited
Streaming	✅ Partial previews	⚠️ Final only
Multi-turn	✅ Full support	⚠️ Basic

🔧 Migration Guide

For New Users

• Default configuration uses Responses API automatically
• No action required - just set OPENAI_API_KEY

For Existing Users

• Backward compatible - existing configurations continue to work
• Optional upgrade: Set API_MODE=responses to use latest features
• Gradual migration supported with dual-mode architecture

📦 Installation

# Clone and install
git clone https://github.com/pavelsukhachev/mcp-server-gpt-image.git
cd mcp-server-gpt-image
npm install

# Configure (uses Responses API by default)
echo "OPENAI_API_KEY=your-api-key-here" > .env

# Run
npm run start:http

🚀 What's Next

This release establishes MCP Server GPT Image as the most advanced OpenAI image generation server available, supporting both cutting-edge Responses API and legacy Images API with seamless switching.

Perfect for:

• 🎨 Advanced image generation workflows
• 🔄 Multi-turn conversation contexts
• ⚡ Real-time streaming applications
• 🧪 Development and production environments

---

Full Changelog: https://github.com/pavelsukhachev/mcp-server-gpt-image/blob/main/CHANGELOG.md

v1.1.0 - Streaming, Caching & Optimization

🚀 New Features

⚡ Real-time Streaming Support

• Server-Sent Events (SSE) endpoint at /mcp/stream
• Live progress updates during image generation
• Partial image previews (1-3 configurable)
• Seamless error handling in stream

💾 Intelligent Caching System

• Two-tier caching (memory + disk)
• Content-based cache keys using SHA256
• Configurable TTL and size limits
• New tools: clear_cache and cache_stats
• Automatic cleanup when limits exceeded

🖼️ Image Optimization Engine

• Automatic compression with Sharp library
• Support for JPEG, PNG, and WebP formats
• Configurable quality/compression levels
• Typical size reductions of 30-80%
• Preserves transparency when needed

🔧 Enhanced Configuration

• New environment variables for cache control
• output_compression parameter for quality
• partialImages parameter for streaming
• format parameter for output format

📝 Documentation

• Comprehensive API documentation
• Environment configuration guide
• Updated examples with streaming demos
• Complete changelog

🐛 Bug Fixes

• Removed unsupported response_format parameter
• Improved error handling for API failures
• Better base64 encoding for large images
• Fixed package-lock.json sync issues

📦 Installation

npm install mcp-server-gpt-image@1.1.0

Or clone and build:

git clone https://github.com/pavelsukhachev/mcp-server-gpt-image.git
cd mcp-server-gpt-image
npm install
npm run build

🏃 Quick Start

Claude Desktop

Add to your Claude Desktop config:

{
  "mcpServers": {
    "gpt-image": {
      "command": "node",
      "args": ["/path/to/mcp-server-gpt-image/dist/index.js", "stdio"],
      "env": {
        "OPENAI_API_KEY": "your-openai-api-key-here"
      }
    }
  }
}

HTTP Server

npm run start:http

See the README for full documentation.

MCP Server GPT Image v1.0.0

🎉 Initial Release

We're excited to announce the first release of MCP Server GPT Image - a Model Context Protocol server that provides access to OpenAI's GPT Image-1 model for advanced image generation and editing.

✨ Features

• 🎨 Image Generation: Create stunning images from text descriptions
• ✏️ Image Editing: Modify existing images with text prompts and masks
• 🔄 Multiple Transports: Support for both stdio and HTTP transports
• 🐳 Docker Support: Easy deployment with Docker and Docker Compose
• 📦 Production Ready: Session management, error handling, and health checks
• 🔒 Secure: Environment-based API key management

🚀 Quick Start

1. Clone the repository
2. Set your OpenAI API key in .env
3. Run npm install && npm run build
4. Use with Claude Desktop or run as HTTP server

📖 Documentation

• Installation Guide
• Available Tools
• Examples
• Docker Setup

🐛 Known Issues

• Streaming image generation is not yet available (pending Responses API support)
• Some advanced editing features require specific mask formats

🙏 Acknowledgments

Built with the Model Context Protocol SDK and powered by OpenAI's GPT Image-1.

---

Full Changelog: https://github.com/pavelsukhachev/mcp-server-gpt-image/commits/v1.0.0