Neura Market
Langchain Chatchat logo

Langchain Chatchat

Free

Langchain-Chatchat（原Langchain-ChatGLM）基于 Langchain 与 ChatGLM, Qwen 与 Llama 等语言模型的 RAG 与 Agent 应用 | Langchain-Chatchat (formerly langchain-ChatGLM), local knowledge based LLM (like ChatGLM, Qwen and Llama) RAG and Agent app with langchain

FreeFree tier
Inputs: text, file, imageOutputs: text, image
Type
Open Source
Visit Website GitHub (38.1k stars)

About Langchain Chatchat

Langchain-Chatchat is an open-source, offline-deployable RAG (Retrieval-Augmented Generation) and Agent application built on Langchain and multiple large language models such as ChatGLM, Qwen, and Llama. It provides a local knowledge base question-answering system that runs entirely with open-source models, supporting both local model deployment frameworks (Xinference, LocalAI, Ollama, FastChat) and online APIs via One API. Key functionalities include LLM dialogue, knowledge base dialogue, search engine dialogue, file dialogue (File RAG with BM25+KNN retrieval), database dialogue, multimodal image dialogue (with models like qwen-vl-chat), ARXIV literature dialogue, Wolfram dialogue, and text-to-image generation. The system offers a Streamlit-based WebUI for multi-session chat and custom system prompts, as well as a FastAPI-based API service. It supports a wide range of model types (LLM, Embedding, Rerank, Vision, Audio) and vector databases, and is designed for Chinese scenarios and offline private deployment.

Key Features

RAG (Retrieval-Augmented Generation) with local knowledge base
Agent capability optimized for ChatGLM3 and Qwen models
Supports multiple model deployment frameworks: Xinference, LocalAI, Ollama, FastChat
Offline private deployment with full open-source stack
WebUI (Streamlit) with multi-session support and custom system prompts
API service based on FastAPI, compatible with OpenAI SDK
File RAG with BM25+KNN and other retrieval methods
Multimodal image dialogue (with models like qwen-vl-chat)
Database dialogue, ARXIV literature dialogue, Wolfram dialogue
Text-to-image generation

Pros & Cons

Pros
  • Fully open-source and free, supports offline deployment
  • Supports a wide range of popular open-source LLMs (ChatGLM, Qwen, Llama, etc.)
  • RAG and Agent capabilities integrated into a single platform
  • Active development with over 38,000 GitHub stars and strong community
  • Comprehensive documentation and multiple deployment options (pip, Docker, source)
  • Extensible via plugin-like model frameworks and online API support
Cons
  • Does not include built-in fine-tuning or training functionality
  • Requires technical expertise for setup and configuration
  • May need powerful hardware (GPU) for optimal performance with large models
  • Some features (Agent, multimodal) are still in active development and may be experimental
  • Primarily focused on Chinese language scenarios, English support may be less mature

Best For

Local knowledge base question-answering for enterprises or organizationsOffline private deployment of LLM applications in sensitive environmentsMulti-turn dialogue with context and document retrievalAutomated tool usage and agent-based task execution (e.g., search, calculation)Multimodal interaction including image understanding and generationAcademic research literature Q&A (ARXIV)

FAQ

Does Langchain-Chatchat support fine-tuning of models?
No, the project does not include fine-tuning or training processes. However, fine-tuning or training external models can be used to optimize the performance of the application.
What LLMs and frameworks are supported?
It supports ChatGLM, Qwen, Llama, and many other LLMs through frameworks like Xinference, LocalAI, Ollama, and FastChat. Online APIs (OpenAI, Azure, Claude, etc.) are also supported via One API.
Can I deploy Langchain-Chatchat offline?
Yes, the project is designed for offline private deployment using entirely open-source models. Supported hardware includes CPU, GPU, NPU, and MPS.