LangChain
The most widely used framework for building LLM applications, with a huge set of integrations for models, vector stores, and tools. It is a common backbone for RAG systems, and works with local models through Ollama and llama.cpp.
Strengths
- Enormous ecosystem of integrations and examples
- Works with local models and local vector stores out of the box
- Pairs with LangGraph when you need stateful, controllable agents
Weaknesses
- The breadth and layers of abstraction can feel heavy for simple tasks
- Frequent API changes have historically made upgrades painful
- Easy to over-engineer a pipeline that a few functions would handle
At a glance
- Licence
- MIT
- Pricing
- Open core
- Platforms
- macOS, Windows, Linux
- GPU required
- No
Links
Entry last verified 30 July 2026.