This project provides a self-hostable, customizable web interface for interacting with OpenAI's GPT models, primarily targeting users who want a private ChatGPT experience. It offers easy deployment options and extensive configuration for tailoring AI behavior and network settings.

How It Works

The application is built using Go and acts as a backend service that communicates with the OpenAI API. It supports proxying requests to bypass network restrictions and allows for detailed configuration of AI parameters like temperature, max_tokens, and presence_penalty to influence response generation. Users can define the AI's persona and behavior through a bot_desc parameter.

Quick Start & Requirements

• Binary Deployment: Download release archives for your OS, copy config.dev.json to config.json, update api_key, and run the executable.
• Source Deployment: git clone, copy config.dev.json to config.json, update configuration, and run with go run main.go.
• Docker: Use docker run with environment variables or mount a configuration file. docker-compose up -d is also supported.
• Prerequisites: OpenAI API key. Optional proxy configuration for network access.
• Documentation: https://github.com/869413421/chatgpt-web

Highlighted Details

• Supports GPT-3.5 and GPT-4 models.
• Includes proxy support (HTTP/SOCKS5) for network access.
• Offers basic authentication and Nginx reverse proxy examples for enhanced security.
• Allows extensive AI personality and behavior customization via configuration parameters.

Maintenance & Community

The project has seen recent fixes and feature additions, including GPT-4 support and user authentication. Community interaction channels are not explicitly mentioned in the README.

Licensing & Compatibility

The project explicitly states "代码仅用于演示和测试" (code is for demo and testing only) and "请勿将本系统代码用于商业用途！" (Please do not use this system code for commercial purposes!). This strongly suggests a non-commercial license, despite no specific license being listed.

Limitations & Caveats

The project is strictly for demonstration and testing purposes and prohibits commercial use. The README warns that using the service provider's API URL means they can see your plaintext requests, including your OpenAI key, recommending self-hosting or trusted sources for the API endpoint.