What it does
The LlamaIndex Framework is an open-source toolkit from LlamaIndex, the company behind LlamaParse, for building agentic applications and retrieval-augmented generation apps. It offers data connectors for APIs, PDFs, documents and SQL, ways to structure data as indices and graphs, and a retrieval and query interface that returns knowledge-augmented answers for an LLM prompt. In Python you can start with the llama-index starter package, or install llama-index-core and add only the integrations you need. The README says there are over 300 integration packages for LLM, embedding and vector store providers, and its examples use OpenAI and models hosted through Ollama. A high-level API lets beginners ingest and query data in a few lines, while lower-level APIs let you customize connectors, indices, retrievers, query engines and reranking. The README notes that the company's main focus has shifted to LlamaParse, an enterprise platform for document parsing, extraction, indexing and document agents, while the open-source framework stays available. LlamaParse requires signing up, and its pricing is not stated on this page.
Capabilities
- Data connectors for APIs, PDFs, documents and SQL
- Indices and graphs that structure data for LLMs
- Retrieval and query interface over your own data
- Over 300 integration packages for LLM, embedding and vector store providers
- High-level API for quick starts, lower-level APIs for customizing
- Works with hosted models such as OpenAI or local ones through Ollama
When to use it
- Building a question-answering app over private documents
- Prototyping a retrieval workflow, then customizing retrievers and rerankers
- Swapping LLM, embedding or vector store providers through integration packages
- Pairing the framework with LlamaParse for hard-to-parse documents
