LlamaIndex
A Python framework focused on RAG and data-augmented LLM applications — provides indexing, retrieval, and query pipelines for connecting LLMs to your data.
LlamaIndex (originally GPT Index) is the leading framework for RAG and structured data + LLM applications. Where LangGraph is agent-first, LlamaIndex is retrieval-first: how do I get my data into the LLM efficiently, with the right context, at the right time.
Strengths: best-in-class document parsing, advanced retrieval strategies (hybrid search, reranking, recursive retrieval), structured output extraction, and SQL-over-data with LLMs. For RAG-heavy applications, LlamaIndex is often the right starting point.
In 2026, LlamaIndex has expanded into agent territory with LlamaIndex Workflows, competing with LangGraph. For pure RAG: LlamaIndex. For complex stateful agents: LangGraph. Many production stacks use both.
Frequently asked
LlamaIndex vs LangChain?+
LlamaIndex is RAG-first; LangChain (LangGraph) is agent-first. For data-heavy applications with retrieval at the core, LlamaIndex wins. For complex multi-step agent work, LangGraph is the better choice.
Is LlamaIndex open source?+
Yes, MIT-licensed. Hosted version (LlamaCloud) exists for enterprise customers wanting managed deployment.