ArchestackBest of Open-Source AI: List on GitHub ↗ (opens in a new tab)

#2525th of 39 in RAG, knowledge and searchFor science, health and research, research and papers

paper-qa

Python RAG library and CLI for answering questions over scientific PDFs with citations

Website ↗ (opens in a new tab)Repository on GitHub ↗ (opens in a new tab)
Stars
9.4k
License
Apache-2.0
Last commit
Aug 2026
Last release
Aug 2026
Language
Python

Missed the latest GitHub check; these facts may be older.

Overview

PaperQA2 indexes local PDFs, text, Office documents and source files, fetches paper metadata from Crossref and Semantic Scholar, and answers questions with in-text citations. An LLM agent runs paper search, evidence gathering and answer generation, using LLM re-ranking and contextual summarization. It ships a `pqa` CLI and a Python API, and works with any LiteLLM-supported model.

Who it is for: Researchers and developers doing cited Q&A over scientific paper collections

Strengths

  • Answers carry in-text citations to specific papers and page ranges
  • Pulls citation counts and retraction checks from Crossref and Semantic Scholar
  • Any LiteLLM provider works, including local servers and Sentence Transformers embeddings
  • Bundled settings (fast, high_quality, wikicrow, contracrow) and a CLI with persistent index

Weaknesses

  • Defaults to OpenAI embeddings and models, so an API key is needed out of the box
  • CalVer since December 2025 drops backwards-compatibility guarantees between releases
  • Library and CLI only, with no web UI or multi-user features
  • Indexing 100+ papers needs Crossref and Semantic Scholar keys to avoid rate limits

What it needs

  • GPU optional
  • Needs LLM provider API key or local LLM server, Crossref API, Semantic Scholar API
  • Models: LiteLLM providers, OpenAI, llamafile, Sentence Transformers embeddings

Also in RAG, knowledge and search

See all 39
Also in RAG, knowledge and search
RankProjectScore
1deeptutorSelf-hosted AI tutor with knowledge bases, guided learning, books and quizzes42k starsApache-2.082 out of 100
2LightRAGGraph-based RAG framework with a REST API server and web UI40.3k starsMIT78 out of 100
3MorphicAI answer engine with cited search results and generative UI components9.2k starsApache-2.077 out of 100
#4MaxKBEnterprise knowledge-base agent platform with RAG, workflows and MCP tools23k starsGPL-3.075 out of 100
#5RAGFlowRAG engine with deep document parsing, template chunking and agentic retrieval Live demo ↗ (opens in a new tab)92.6k starsApache-2.074 out of 100