Why it matters
- 200M+ paper index covers all scientific disciplines — one search engine for medicine, computer science, physics, social sciences, and more.
- Semantic search understands research concepts rather than keywords — finding "few-shot learning" papers without exact terminology match.
- Citation influence scoring helps identify the most impactful papers, not just the most cited — crucial for literature reviews.
- Free public API enables building research tools, literature review automation, and citation analysis pipelines without scraping.
Key capabilities
- Semantic search: Concept-based search that understands research ideas beyond exact keyword matching.
- Paper TLDRs: AI-generated one-sentence paper summaries for quick relevance assessment.
- Citation graphs: Visualize how papers cite each other and trace the development of research ideas.
- Influence scores: Measure paper impact beyond raw citation counts — accounts for citing paper quality.
- Research alerts: Set up email alerts for new papers on specific topics or by specific authors.
- Author profiles: Follow researchers and track their publication history.
- Free API: REST API for programmatic access to 200M+ papers, citations, and search.
- Research libraries: Save papers into organized collections and share with collaborators.
Technical notes
- Index size: 200M+ papers across all scientific disciplines
- AI features: TLDR generation model (open source); semantic search via embeddings
- API: Free REST API;
pip install semanticscholarPython wrapper - Coverage: All fields — CS, medicine, biology, physics, economics, social sciences
- Creator: Allen Institute for AI (AI2); Seattle; non-profit research organization
- Funding: Paul G. Allen Foundation; Microsoft; non-profit/grant funded
- Cost: Completely free; API free for research use
Ideal for
- Researchers conducting systematic literature reviews who need comprehensive coverage and citation analysis.
- Developers building academic research tools, citation analyzers, or paper recommendation systems using the free API.
- Anyone following AI/ML research who wants to quickly assess paper relevance via TLDRs and semantic search.
Not ideal for
- Research that requires AI-synthesized answers to questions across papers — Elicit or Consensus provide that layer.
- Non-English literature — coverage is predominantly English-language publications.
- Real-time preprint tracking — arXiv search is faster for very recent AI/ML work.