LangChain
Framework for developing applications powered by large language models, agents and retrieval-augmented generation.
Overview
LangChain is a framework for building LLM-powered applications and agents by chaining together interoperable components - chat models, embeddings, vector stores, retrievers, and third-party integrations - behind a standard interface, so you can swap providers as the underlying model landscape changes without rewriting application logic.
It's designed to scale from quick prototypes to more complex systems: start with LangChain's high-level abstractions, and move to lower-level components, or to companion tools like LangGraph (for building more controllable, stateful agent workflows) and LangSmith (for evaluating, debugging, and deploying LLM applications), as requirements grow. It's available for both Python and JavaScript/TypeScript.
LangChain is a fit for developers building retrieval-augmented, tool-using, or multi-step LLM applications who want a common set of abstractions and a large library of integrations, rather than writing glue code for each model or vector store from scratch.
- Categories
- AI & Machine Learning
- Languages
- Python
- License
- MIT
Related Projects
Ollama
MITRun large language models locally with a simple command-line interface and API.
- AI & Machine Learning
CAMEL
Apache-2.0Open-source framework for research and building systems of multiple communicating, role-playing AI agents that collaborate autonomously.
- AI & Machine Learning
CrewAI
MITMulti-agent orchestration framework for building teams of collaborating, role-based AI agents that work together autonomously to complete complex tasks.
- AI & Machine Learning
Docling
MITDocument parsing library that converts PDFs, Word docs, and other formats into clean, structured data ready for LLM and RAG pipelines.
- AI & Machine Learning
Spotted an error? Suggest an edit on GitHub.