PageIndex
📑 PageIndex: Document Index for Vectorless, Reasoning-based RAG
What is PageIndex?
📑 PageIndex: Document Index for Vectorless, Reasoning-based RAG
Overview of PageIndex
What it is
PageIndex is a vectorless, reasoning-based RAG engine. It replaces the vector index with a hierarchical tree index and uses an LLM to reason through it for retrieval. The process involves two steps: generating a tree-structure index for each document, then agentically searching that tree with LLM reasoning. It does not require vector databases or chunking. According to the README, PageIndex reached 98.7% accuracy on the FinanceBench financial document QA benchmark.
Key features
- No vector database or chunking required - Uses a hierarchical tree index instead of a vector index - Retrieval via LLM reasoning over the tree, enabling context-aware and traceable results - Supports local and cloud modes (PageIndex Cloud for scanned/image-rich documents)
- Page Search API can cut token cost by 97% on document question answering (as stated in the README)
- Can integrate with OpenAI Agents SDK, Claude Agent SDK, or other frameworks
Use cases
PageIndex is designed for long, complex professional documents such as financial reports, legal documents, regulatory filings, technical manuals, medical literature, and academic textbooks.
Getting started
bash pip install -U pageindex python import os from pageindex import PageIndexClient os.environ['OPENAI_API_KEY'] = 'your-openai-key' client = PageIndexClient( index='gpt-6-luna', chat='gpt-6-sol', ) doc_id = client.submit_document('report.pdf')['doc_id'] answer = client.chat('What was the 2023 operating margin?', doc_id=doc_id) print(answer)
License and pricing
The project is open source under the MIT license. PageIndex Cloud is a separate service; pricing for dedicated deployment (VPC or on-premises) is not specified in the README.
Based on the project's public README and repository metadata.