#77th of 17 in RAG and knowledge
pageindex
Vectorless RAG that builds a tree index of a PDF and lets an LLM search it
Live demo ↗
(opens in a new tab)Documentation ↗
(opens in a new tab)Website ↗
(opens in a new tab)Repository on GitHub ↗
(opens in a new tab)
- Stars
- 39.1k
- License
- MIT
- Last commit
- Oct 2026
- Last release
- Oct 2026
- Language
- Python
22,427 on 24 Mar 202639,113 on 13 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
PageIndex builds a hierarchical tree index from each PDF, then has an LLM reason through that tree to find the relevant sections, with no vector database or chunking. The Python SDK (pip install pageindex) runs locally with your own LLM key for text-based PDFs, or points at PageIndex Cloud for OCR, scanned documents and block-level citations. The README reports 98.7% on FinanceBench and about $0.001 per page to index locally.
Who it is for: Developers building document Q&A over long PDFs such as filings and manuals
Strengths
- No vector DB or chunking; retrieval is traceable to page-level references
- Local mode indexes a 1,000-page PDF for roughly a dollar, per README benchmarks
- Same client switches between local and Cloud by changing the index argument
- Documents integration with OpenAI Agents SDK and Claude Agent SDK
Weaknesses
- Local mode handles text-based PDFs only; no OCR or image understanding
- MCP server, metadata, folders and File System for large corpora are Cloud-only
- Each query needs LLM reasoning calls, so accuracy and cost depend on the chat model
- Needs an external LLM API key; local-model support is not stated in the README
What it needs
- no GPU
- Needs LLM API key (e.g. OpenAI), PageIndex API key (Cloud mode only)
- Models: OpenAI models (gpt series), other providers configurable per docs
Also in RAG and knowledge
See all 17| Rank | Project | Score |
|---|---|---|
| 1 | LightRAGGraph-based RAG framework with a REST API server and web UI | 75 out of 100 |
| 2 | Open NotebookSelf-hosted NotebookLM alternative with podcasts and 20+ model providers | 74 out of 100 |
| 3 | RAGFlowRAG engine with document parsing, agentic retrieval and citations | 73 out of 100 |
| #4 | MaxKBEnterprise knowledge-base agent platform with RAG, workflows and MCP tools | 71 out of 100 |
| #5 | SurfSenseDesktop agent that researches and edits your local documents with cited answers | 68 out of 100 |