ChuanhuChatGPT

ChuanhuChatGPT is a self-hosted web chat interface for people who want local models and cloud AI services in the same application. It runs on your computer or server and opens in a browser, with support for Windows, macOS and Linux. The Python application is open source under GPL-3.0.

Local model support includes ChatGLM, LLaMA with LoRA models, Qwen, StableLM and MOSS. It can also connect to custom local inference services. Cloud connections include OpenAI's GPT models, Azure OpenAI, Claude, Google Gemini and DeepSeek R1. Hosting the interface yourself doesn't make those API calls local: cloud models process requests through their providers, while local model inference runs on your own hardware.

Beyond chat, ChuanhuChatGPT can answer questions about uploaded files through a knowledge base and use web search to bring online information into responses. Its agent assistant handles tasks automatically in an approach similar to AutoGPT. It also supports GPT-3.5 fine-tuning and image generation through DALL·E 3.

Chats are saved automatically, with separate histories for each user. You can search and rename conversations, or let a model name them. Prompt templates and system prompts support recurring tasks and role-based conversations, and responses can include rendered equations, tables and highlighted code.

The interface adapts to phones and supports installation as a PWA through Chrome, Edge and Safari. A hosted browser demo is available on Hugging Face Spaces.