Skip to main content
PromptQuorumBuilt for humans. Structured for AI.
Home/Power Local LLM/Langchain-Chatchat Review: Self-Hosted RAG and Agent Framework
RAG & Document Chat

Langchain-Chatchat Review: Self-Hosted RAG and Agent Framework

·11 min read·By Hans Kuepper · Founder of PromptQuorum · Discovery engine for open-weight & open-source AI
Read in:
🇺🇸en🇩🇪de🇫🇷fr🇯🇵ja🇨🇳zh🇪🇸es🇧🇷pt🇸🇦ar🇰🇷ko

Langchain-Chatchat (formerly langchain-ChatGLM) is a free, open-source, self-hosted RAG and Agent application built on the LangChain framework, letting you run local knowledge-base question-answering against models like ChatGLM, Qwen, and Llama, or against cloud APIs. Developed by the chatchat-space GitHub organization and licensed under Apache-2.0, it has no paid tier — you install it yourself via pip or Docker and run it on your own hardware or server, choosing which model backend to connect. The project has over 38,600 GitHub stars, though its release cadence has slowed since its most recent tagged version, v0.3.1, shipped in July 2024.

Langchain-Chatchat (github.com/chatchat-space/Langchain-Chatchat) is a free, open-source, self-hosted RAG and Agent application built on the LangChain framework, formerly named langchain-ChatGLM before its August 2023 rename. It runs local knowledge-base question-answering against models such as ChatGLM, Qwen, and Llama through pluggable backends, or against cloud APIs, and has over 38,600 GitHub stars. This review covers what it does, its version history, how to install it, and how it compares to other self-hosted RAG tools.

Key Takeaways

  • Langchain-Chatchat (github.com/chatchat-space/Langchain-Chatchat) is a free, open-source, self-hosted RAG and Agent application built on the LangChain framework — not a downloadable end-user app
  • Formerly named langchain-ChatGLM; renamed to Langchain-Chatchat in August 2023 as its model support broadened beyond ChatGLM
  • Apache-2.0 licensed, confirmed via the GitHub repository's LICENSE file
  • Supports ChatGLM, Qwen, Llama, and other models through pluggable backends — Xinference, Ollama, LocalAI, FastChat, One API — or cloud APIs
  • Over 38,600 GitHub stars and 6,200+ forks, 26 open issues, as of this review (2026-09-18); the repository is not archived
  • Latest tagged GitHub release is v0.3.1 (July 12, 2024); most recent repository push was November 10, 2025 — a real gap between the newest commit activity and the latest version tag

📍 In One Sentence

Langchain-Chatchat is a free, open-source, self-hosted RAG and Agent framework built on LangChain, formerly named langchain-ChatGLM, that runs local knowledge-base question-answering against models like ChatGLM, Qwen, and Llama, or against cloud APIs, with no paid tier.

💬 In Plain Terms

Instead of paying for a hosted document-chat service, Langchain-Chatchat lets you install a free tool on your own server that reads your documents, breaks them into searchable chunks, and answers questions about them using an AI model you choose — a local one running on your own hardware, or a cloud API you connect with your own key.

📌Note: This review is the companion to Langchain-Chatchat's entry in the Local LLM Software Directory — see that page for how it compares at a glance to dozens of other local AI tools.

What Is Langchain-Chatchat?

Langchain-Chatchat is a free, open-source, self-hosted application for local knowledge-base retrieval-augmented generation (RAG) and Agent workflows, built on the LangChain framework. Its GitHub repository description states it is a RAG and Agent application based on LangChain and models including ChatGLM, Qwen, and Llama. It was originally released as langchain-ChatGLM around April 2023, built specifically around the ChatGLM-6B model, and renamed to Langchain-Chatchat in August 2023 as its model support broadened.

  • Product type: a self-hosted web application (Streamlit-based UI) plus a Python package and Docker image — not a downloadable end-user binary
  • Developer: the chatchat-space GitHub organization
  • Repository: github.com/chatchat-space/Langchain-Chatchat, created March 31, 2023
  • License: Apache-2.0, confirmed via the repository's LICENSE file
  • Scale: over 38,600 GitHub stars and 6,200+ forks, 26 open issues, as of this review (2026-09-18); the repository is not archived
  • Origin: a Chinese-origin open-source project with a bilingual (Chinese/English) README, originally built around the ChatGLM model family and now supporting Qwen, Llama, and other models via pluggable backends

Langchain-Chatchat's Project History and Version Milestones

Langchain-Chatchat has shipped through several architecture changes since its original ChatGLM-6B-only release in 2023, though its release cadence has slowed markedly since mid-2024.

  1. 1
    v0.1.0 — around April 2023: Original release as langchain-ChatGLM
    Why it matters: The project's starting point: a single-model, ChatGLM-6B-specific local knowledge-base question-answering application.
  2. 2
    Renamed to Langchain-Chatchat — August 2023, with v0.2.0 shipping FastChat integration
    Why it matters: Marked the project's shift from a ChatGLM-only tool to a broader, multi-model-backend framework.
  3. 3
    v0.2.5 — October 2023: Added Agent features
    Why it matters: Introduced agent-style, tool-using workflows alongside the original RAG question-answering pipeline.
  4. 4
    v0.2.8 — December 1, 2023
    Why it matters: Incremental tagged release in the v0.2.x line.
  5. 5
    v0.2.9 — December 31, 2023
    Why it matters: Incremental tagged release closing out 2023.
  6. 6
    v0.2.10 — January 25, 2024
    Why it matters: Final release in the v0.2.x line before the v0.3.0 architecture change.
  7. 7
    v0.3.0 — June 20, 2024: Described in the project's own changelog as a "new architecture"
    Why it matters: A significant internal rework of the application's architecture.
  8. 8
    v0.3.1 — July 12, 2024: The most recent tagged GitHub release found by this review
    Why it matters: This is the newest version tag as of this review — no newer tagged release has been found.

What Does Langchain-Chatchat Actually Do?

Langchain-Chatchat's core feature is a local knowledge-base RAG pipeline: it splits your documents into chunks, converts them into vector embeddings, retrieves the most relevant chunks for a given question by similarity search, and passes them to an LLM to generate an answer.

  • Local knowledge-base RAG pipeline: text splitting → vectorization → similarity matching → LLM response generation
  • Agent support, particularly optimized for ChatGLM3 and Qwen models, for tool-using and multi-step workflows
  • Multiple pluggable model-serving backends: Xinference, Ollama, LocalAI, FastChat, and "One API" integration, so you can choose which inference server runs your model
  • Multimodal support: image dialogue and text-to-image capabilities
  • Database/SQL dialogue ("chat with your database") functionality for querying structured data in natural language
  • Web UI built with Streamlit, supporting multiple simultaneous conversations

Usage Examples: Three Ways to Use Langchain-Chatchat

These are concrete workflows built from Langchain-Chatchat's documented features above — not hypothetical use cases.

Build a local knowledge base and query it

Upload a set of documents (PDFs, text files, and similar) into Langchain-Chatchat's knowledge-base interface. The app splits them into chunks, generates vector embeddings, and stores them for retrieval. Once indexed, ask natural-language questions and the app retrieves the most relevant chunks before generating an answer with your connected model — a self-hosted equivalent of uploading files to a cloud document-chat service.

Connect a local model backend like Ollama or Xinference

Rather than calling a cloud API, point Langchain-Chatchat at a local inference server such as Ollama or Xinference running ChatGLM, Qwen, or Llama. The app routes its RAG and Agent requests through that backend, so document content and prompts stay on hardware you control.

Use the SQL/database chat feature

Connect Langchain-Chatchat to a structured database and use its chat-with-your-database functionality to ask questions in natural language instead of writing SQL directly — useful for exploring a dataset without switching to a separate query tool.

Is Langchain-Chatchat Free?

Yes — Langchain-Chatchat is free and open source, with no separate paid tier found by this review. It is licensed under Apache-2.0, confirmed via the repository's LICENSE file, and this review found no vendor-run hosted/cloud version of the tool — you install and run it yourself.

  • No subscription, license fee, or usage cap imposed by Langchain-Chatchat itself
  • Self-hosted: you provide your own server/hardware and pay only for that infrastructure, plus any cloud model API costs if you choose a cloud backend instead of a local model
  • Apache-2.0 license permits commercial use, modification, and redistribution, subject to the license's standard attribution and notice terms

Langchain-Chatchat vs. RAGFlow

Langchain-Chatchat and RAGFlow are both self-hosted, open-source RAG applications, but they take different approaches to how they process and answer questions about your documents.

Primary focus

Langchain-Chatchat:
Local knowledge-base RAG and Agent workflows built directly on the LangChain framework
RAGFlow:
Document understanding-focused RAG — see the dedicated RAGFlow review for verified details

Model backend flexibility

Langchain-Chatchat:
Pluggable backends: Xinference, Ollama, LocalAI, FastChat, One API, or cloud APIs
RAGFlow:
See the RAGFlow review for its own backend support

Multimodal support

Langchain-Chatchat:
Image dialogue and text-to-image capabilities
RAGFlow:
See the RAGFlow review for details

SQL/database chat

Langchain-Chatchat:
Yes — dedicated chat-with-your-database functionality
RAGFlow:
See the RAGFlow review for details

Maintenance cadence

Langchain-Chatchat:
Latest tagged release v0.3.1 (July 2024); most recent repository push November 2025
RAGFlow:
See the RAGFlow review for its current release cadence

License

Langchain-Chatchat:
Apache-2.0
RAGFlow:
See the RAGFlow review for its license

This table reflects Langchain-Chatchat's own verified facts; for RAGFlow's specific claims, read the RAGFlow review directly rather than relying on this summary alone.

Compare in directory

Who Should Use Langchain-Chatchat?

Whether Langchain-Chatchat fits depends on whether you want a free, self-hosted RAG and Agent framework built on LangChain, and whether its slower recent release cadence is acceptable for your use case.

Use Langchain-Chatchat if you

  • Want a free, self-hosted RAG and Agent framework built directly on LangChain, rather than a hosted SaaS product
  • Are building a knowledge-base Q&A app for Chinese-language or bilingual (Chinese/English) content and want a tool with ChatGLM/Qwen-optimized Agent support
  • Want to choose your own model backend (Xinference, Ollama, LocalAI, FastChat, or a cloud API) rather than being locked into one
  • Need database/SQL chat alongside document RAG in the same application
  • Are comfortable deploying and maintaining a self-hosted Python/Docker application yourself, including its dependencies

Look elsewhere if you

  • Want a project with a frequent, recent release cadence — this review found a roughly 16-month gap between Langchain-Chatchat's latest tagged release (v0.3.1, July 2024) and its most recent repository push (November 2025)
  • Need a downloadable end-user desktop app rather than a self-hosted web application you deploy yourself
  • Want built-in enterprise support, an SLA, or a vendor-run managed cloud version — this review found no such offering
  • Need a RAG tool specifically focused on deep, layout-aware document understanding — see the RAGFlow review for a comparison

Competitors and Alternatives

Langchain-Chatchat sits in the self-hosted RAG and document-chat segment alongside RAGFlow, LlamaIndex, Haystack, and MaxKB.

Haystack (deepset)

Best known for:
Open-source RAG and search-pipeline framework from deepset
Link:
Haystack review

This list reflects tools commonly compared to Langchain-Chatchat in the self-hosted RAG/document-chat segment, not an independent PromptQuorum ranking — see the Local LLM Software Directory for the full catalog, and verify each tool's current feature set and license before choosing.

Compare in directory

Common Mistakes When Evaluating Langchain-Chatchat

Most confusion about Langchain-Chatchat comes from its naming history, assumptions about its release cadence, or its relationship to the LangChain framework it is built on.

Mistake 1: Assuming it's still ChatGLM-only under its old name

The project was originally released as langchain-ChatGLM around April 2023, built specifically around the ChatGLM-6B model. It was renamed to Langchain-Chatchat in August 2023 and now supports multiple model backends beyond ChatGLM, including Qwen and Llama, via pluggable serving layers.

Mistake 2: Assuming frequent recent releases

Langchain-Chatchat's most recent tagged GitHub release is v0.3.1, published July 12, 2024. This review found the repository's most recent push dated November 10, 2025 — ongoing commit activity, but not matched by a corresponding recent version tag. Verify the current release and commit status directly on GitHub rather than assuming a fast release cadence.

Mistake 3: Confusing it with plain LangChain

Langchain-Chatchat is built on top of the LangChain framework but is a separate, standalone application with its own repository, UI, and Agent/RAG feature set — it is not the same project as the general-purpose LangChain library itself.

Mistake 4: Assuming it's cloud-only or local-only

Langchain-Chatchat is hybrid: it supports local model backends (Xinference, Ollama, LocalAI, FastChat) for fully self-hosted inference, as well as cloud API connections, depending on how you configure it — neither mode is enforced by the software itself.

Frequently Asked Questions

What is Langchain-Chatchat?

Langchain-Chatchat (github.com/chatchat-space/Langchain-Chatchat) is a free, open-source, self-hosted RAG and Agent application built on the LangChain framework, formerly named langchain-ChatGLM before its August 2023 rename.

Is Langchain-Chatchat free?

Yes. It is licensed under Apache-2.0, and this review found no separate paid tier, managed cloud offering, or commercial support plan.

How do I install Langchain-Chatchat?

Via pip (pip install langchain-chatchat -U, or `pip install "langchain-chatchat[xinference]" -U for the Xinference backend) or Docker (docker pull chatimage/chatchat`). The project's own documentation strongly recommends docker-compose deployment for production use.

What license does Langchain-Chatchat use?

Apache-2.0, confirmed via the GitHub repository's LICENSE file.

What models and backends does Langchain-Chatchat support?

ChatGLM, Qwen, Llama, and other models through pluggable backends — Xinference, Ollama, LocalAI, FastChat, and "One API" integration — or cloud APIs.

Is Langchain-Chatchat actively maintained?

This review found a gap between its most recent tagged release (v0.3.1, July 12, 2024) and its most recent repository push (November 10, 2025), roughly 16 months apart. There is ongoing commit activity, but the release cadence has slowed substantially since mid-2024 — verify current status directly on GitHub.

What is the difference between Langchain-Chatchat and LangChain itself?

Langchain-Chatchat is built on top of the LangChain framework but is a separate, standalone RAG and Agent application with its own repository, UI, and feature set — not the general-purpose LangChain library itself.

Does Langchain-Chatchat support Chinese language and ChatGLM?

Yes. The project was originally built around the ChatGLM model family (under the earlier name langchain-ChatGLM), has a bilingual Chinese/English README, and remains a Chinese-origin open-source project.

Does Langchain-Chatchat support SQL/database chat?

Yes, it includes a dedicated chat-with-your-database functionality for querying structured data in natural language.

Who maintains Langchain-Chatchat?

The chatchat-space GitHub organization maintains the project's canonical repository.

Sources

← Back to Power Local LLM