# LlamaIndex

> A free kit of code that hooks AI up to your own files.

**Field:** Software Development  
**Category:** AI Engineering  
**Also known as:** LlamaIndex Framework, GPT Index

## Think of it like

Think of it like a school librarian. They sort the books, then find the right page when you ask.

## Definition

LlamaIndex is a free, open-source framework for linking language models to your own data. It started in 2022 under the name GPT Index. It has loaders for many kinds of files, such as PDFs, web pages and data stores. It also has tools to split, embed and index what they load. Then a query engine finds the right bits for a question and hands them to the model. It is used most to build RAG apps and agents that work with files.

## Example

Lucy points LlamaIndex at 300 PDF contracts. It loads and indexes them, and her team can now ask which contracts renew in March.

## Why it matters

It does the dull parts of search for you, like reading lots of file types and splitting them well.

## Common mistake

Keeping the same chunk size for every kind of file. Tables, code and long reports each need their own split.

## Three principles

- **Your data first.** Loaders for many sources.
- **Split and index.** Ready for search.
- **Query engine.** Finds the right pieces.

## Related terms

- [LLM Framework](https://learn.benibix.com/term/development/llm-framework)
- [LangChain](https://learn.benibix.com/term/development/langchain)
- [Retrieval-Augmented Generation (RAG)](https://learn.benibix.com/term/ai/retrieval-augmented-generation)
- [Chunking (RAG)](https://learn.benibix.com/term/ai/chunking)

## Told apart from

- [LangChain vs LlamaIndex](https://learn.benibix.com/compare/langchain-vs-llamaindex): People think LangChain and LlamaIndex do the same job, and they do not. LangChain joins steps: a prompt, a model, a tool and a parser, one after another. LlamaIndex gets your own files ready for a model. It loads them, splits them and finds the right parts for each question. Many apps use both.

![Documents go through loaders and are split and indexed. A question then runs through a query engine that finds the right pieces and hands them to the model.](https://learn.benibix.com/diagrams/dev-llamaindex.svg)

---

Source: https://learn.benibix.com/term/development/llamaindex
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-20
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
