Researchers and scientists who need accurate, citation backed answers drawn from large collections of papers.

PaperQA, now PaperQA2, is an agentic retrieval augmented generation system that answers questions over PDFs, text, and Office documents with in text citations and retraction checks.

Runs locally or in the cloud as a Python package, invoked via CLI, library API, or as an agent tool.

First released in 2023 by Future House and substantially upgraded as PaperQA2 through 2025 and 2026.

Reduces hallucination in scientific literature review by grounding every answer in retrieved passages with verifiable citations.

Uses LiteLLM to connect to any supported LLM provider and supports both local and cloud embedding models.

Open source under Apache 2.0 on GitHub, installable via pip, no account required though it needs API keys for whichever LLM provider is configured.