Source: Notebook tutorial: LangChain vs LlamaIndex
The owner's comparison of LangChain (orchestration and agents) and LlamaIndex (data ingestion and retrieval), with the same RAG built both ways.
Contents
- What: interactive HTML tutorial in the owner’s public Notebook
- Author: Loic (coolloic)
- Date: undated (read 2026-11-14)
- Original: LangChain vs LlamaIndex
Summary
Frames LangChain as a broad orchestration toolkit and LlamaIndex as a focused data framework; they overlap on RAG. Builds the same document Q&A in both to show LlamaIndex’s opinionated defaults versus LangChain’s explicit, composable stages.
Key claims
- LangChain’s limits are sprawl: many abstractions, fast-moving APIs, and more manual RAG assembly.
- LlamaIndex reaches a working RAG answer in about four lines because it decides the pipeline; LangChain exposes each seam, which costs code but gives control and reuse inside agents and graphs.
- Rule of thumb: if the hard part is the data, start with LlamaIndex; if it is control flow, start with LangChain/LangGraph; combine them by exposing a LlamaIndex retriever as a tool in a LangGraph agent.