LlamaIndex

framework active open-source

A data framework for LLM applications, specializing in connecting LLMs to external data sources. LlamaIndex provides sophisticated indexing, retrieval, and query engines for RAG applications, with particular strength in complex document structures and multi-step retrieval.

Implements

Concepts this tool claims to implement:

  • Core focus is RAG. VectorStoreIndex, TreeIndex, KeywordTableIndex, and other index types. Query engines with synthesis.

  • Chunking primary

    NodeParser hierarchy with sentence splitter, semantic splitter, hierarchical node parser. Sophisticated chunking strategies.

  • Retriever primary

    Multiple retriever types: vector, keyword, hybrid, recursive, auto- merging. Retriever composition and routing.

  • Document abstraction, node graphs, and index persistence. Knowledge graph index for structured data.

  • Agent abstractions with tool use, query planning, and sub-question decomposition for complex queries.

Integration Surfaces

  • Python SDK
  • TypeScript SDK
  • LlamaCloud (managed service)
  • LlamaParse (document parsing)

Details

Vendor
LlamaIndex Inc.
License
MIT
Runs On
local, cloud
Used By
human, agent, system

Notes

LlamaIndex (formerly GPT Index) is the leading RAG-focused framework. While LangChain is broader, LlamaIndex goes deeper on retrieval with advanced indexing strategies, multi-document reasoning, and structured data handling.