PageIndex
New Active Docs only Last checked 2026-11-16 Permissive (MIT/Apache/BSD) Developer library
๐ฌ One weekly email: local RAG releases, hardware compatibility updates, and practical setup guides. Free. Unsubscribe anytime.
Updates on X: @mkjerman
What it is
PageIndex is a vectorless, reasoning-based RAG engine from Vectify AI. Instead of chunking documents into a vector database, it builds a hierarchical tree index (like a table of contents) for each document and has an LLM reason its way down the tree to the right section, so answers trace back to explicit pages.
The MIT-licensed SDK has a local mode (no PageIndex server or API key) and a paid PageIndex Cloud. Per the README, the open-source version handles text-based PDFs; OCR, scanned documents and the MCP server are Cloud features. The quickstart uses an OpenAI key; the v0.5.1 release notes say keyless OpenAI-compatible servers work too. RAGShelf has not run it. See our vector vs vectorless RAG guide.
- License
- MIT
- Latest release
- v0.7.0
- Release date
- 2026-11-15
- Hardware
- Python library (pip install pageindex); no vector DB; needs an LLM for indexing summaries and tree search unverified
- Document types
- Text-based PDFs in the open-source SDK (scanned/image-rich docs and OCR are Cloud-only per README) unverified
- OCR / tables / retrieval
- No OCR in the open-source version; vectorless tree index built from PDF layout, LLM reasons over the tree unverified
- Cloud calls
- Defaults to OpenAI in the README quickstart (your API key); local mode needs no PageIndex server or key, and v0.5.1 notes list keyless OpenAI-compatible servers unverified
- Fully offline
- Needs config
- GitHub stars
- 40,452
- Last commit
- 2026-11-15 (last push, GitHub API 2026-11-16)
Install
Copied from the official README. Check the project docs for options.
pip install -U pageindexRetrieval
Layout / ToC tree Agentic / iterative
Based on: README: hierarchical tree index, LLM reasoning over the tree; no vector DB or chunking (checked 2026-11-09) (checked 2026-10-30). Vector vs. vectorless RAG explained โ
NVIDIA DGX Spark
DGX Spark: Unknown
No DGX Spark report found yet. We only mark a tool as working when there is a published source.
Last updated: 2026-11-16. License and release data from the GitHub API. Fields marked unverified have not yet been checked against official docs. How we know: Docs only (see the legend). Spot something wrong? Send a correction.
๐ฌ Get next week's tool
One new local RAG tool in your inbox every Monday. Free, unsubscribe anytime.
Sample issue: "AirLLM: run a 70B model on a 4GB GPU by loading one layer at a time. What it's good for (slow, private batch Q&A over your docs), what it isn't (fast chat), and the install command." Plus one new setup recipe or page, like our local vs. cloud cost calculator.