NextChat

NextChat is an open-source desktop app that provides a lightweight front-end for interacting with various large language models via API keys or self-hosted deployments. The software is available as a cross-platform desktop application for Windows, MacOS, and Linux, as well as a web-based version that can be deployed on Vercel.

Users can connect the interface to multiple AI providers or local LLM runners to centralize their AI workflows. The application focuses on speed and privacy, storing conversation data locally in the browser or on the desktop client rather than on a remote server. It is designed for users who want a unified interface to manage different AI models without relying on a single provider's web portal or subscription service.

Key features

• Support for DeepSeek and GPT-4
• Compatibility with self-deployed LLMs such as LocalAI and RWKV-Runner
• Local data storage to ensure user privacy
• Markdown support including LaTex, mermaid diagrams, and code highlighting
• Prompt templates for creating, sharing, and debugging chat tools
• Automatic chat history compression to manage token usage
• Artifacts window for previewing and sharing generated webpages
• Plugin support for network search and calculators
• Multi-language support including English, Chinese, French, and Spanish

The application is built using Next.js and React, with the desktop version utilizing Tauri to maintain a small binary size of approximately 5MB. It integrates with a wide range of API endpoints, including Azure and various cloud providers, and supports streaming responses for faster interaction. For corporate users, an enterprise edition exists that adds brand customization, administrative permission controls, security auditing for sensitive inquiries, and private cloud deployment options to ensure data security.

NextChat serves as a flexible, privacy-centric shell for users who prefer to manage their own AI API keys across different model providers while maintaining control over their conversation history.