NextChat is a lightweight and fast AI assistant application designed for broad accessibility across web, mobile, and desktop platforms. It aims to provide a seamless user experience for interacting with various large language models, including OpenAI's GPT series, Claude, Gemini Pro, and DeepSeek, with a focus on privacy and local data storage.

How It Works

The application is built as a Progressive Web App (PWA) with a compact client size, ensuring fast initial loading. It supports streaming responses and markdown rendering for rich content display. A key architectural choice is its compatibility with self-hosted LLMs via integrations like RWKV-Runner and LocalAI, allowing users to leverage local or private model deployments. Data is stored locally in the browser, enhancing user privacy.

Quick Start & Requirements

• Local Development: Requires NodeJS >= 18 and Yarn. Install dependencies with yarn install and run with yarn dev.
• Docker Deployment: docker pull yidadaa/chatgpt-next-web and docker run -d -p 3000:3000 -e OPENAI_API_KEY=sk-xxxx -e CODE=your-password yidadaa/chatgpt-next-web.
• Prerequisites: An API key for the desired LLM service is required.
• Documentation: [docs][./docs]

Highlighted Details

• Supports multiple LLM providers including OpenAI, Azure, Google Gemini, Anthropic Claude, and more.
• Features prompt templates, chat history compression, and sharing conversations as images or to ShareGPT.
• Extensible with plugins for network search, calculators, and custom APIs.
• Offers an Enterprise Edition for private deployment, brand customization, and permission control.

Maintenance & Community

The project is actively maintained, with recent updates including real-time chat and enhanced security for Tauri. Community support and translation contributions are encouraged.

Licensing & Compatibility

• License: MIT.
• Compatibility: Permissive MIT license allows for commercial use and integration into closed-source projects.

Limitations & Caveats

The project relies on user-provided API keys for most functionalities, and while it supports self-hosted models, setup complexity may vary. Some advanced features like local knowledge base integration are still under development.