llama_index

Project Overview

LlamaIndex is an open-source data framework for building applications with large language models. It ingests data from PDFs, APIs, SQL databases, and documents, then organizes it into indices that LLMs can query. The framework core and 300+ integration packages are published separately, so you can pick your preferred model and vector store providers.

Turn private data into LLM context with one toolkit covering ingestion, indexing, and querying.

Key Features

• Data connectors: ingest data from APIs, PDFs, SQL, documents, and more
• Index construction: organize data into vector, keyword, and other index types for retrieval
• Query engines: retrieve relevant chunks and compose answers automatically
• RAG pipelines: a complete ingestion, chunking, embedding, and generation flow out of the box
• Agent orchestration: build single-agent and multi-agent apps with Workflows
• Modular integrations: LLMs, embedding models, and vector stores plug in as separate packages with low switching cost
• Structured output: constrain answers to JSON and other fixed formats

Target Audience

• Developers building RAG knowledge-base assistants
• Engineers wiring enterprise documents into LLMs
• AI teams developing multi-agent applications
• Product builders prototyping LLM apps quickly

Quick Start

1. Install the starter package (core plus common integrations):

pip install llama-index

2. Or install core only and add integrations as needed:

pip install llama-index-core

3. Follow the official docs to configure an LLM and vector store, build an index, and query it

Imports with core come from the core package; those without come from integration packages, for example:

from llama_index.core.llms import LLM
from llama_index.llms.openai import OpenAI

Core Advantages

• Complete ecosystem: ready-made components across the whole path from ingestion to answers
• Vendor neutral: 300+ integration packages let you swap models and storage backends freely
• Quick start: one command installs the starter, and a few lines run your first query

System Requirements

Notes

• Project focus: the team now emphasizes the LlamaParse document-parsing platform; the framework remains usable but updates follow the docs site
• Package layout: choose either the llama-index starter or llama-index-core, then install integration packages separately
• Docs freshness: the README lags behind the documentation site, so treat the official docs as current