Academic Papers API: OpenAlex, Crossref, arXiv for AI agents

Search scholarly papers across OpenAlex, Crossref and arXiv in a single run and get back one clean record per paper, deduplicated by DOI: title, authors, year, venue, DOI, citation count, open-access status, PDF and landing-page links, and the licence.

Built for AI agents, RAG pipelines, literature reviews, research assistants and academic dashboards that need reliable paper metadata without juggling three APIs, their rate limits and their different formats.

Why use it

• Three open databases, one format. OpenAlex (250M+ works with citations and open-access data), Crossref (the DOI registry) and arXiv (preprints) come back in the same fields.
• No duplicates. The same paper found in two or three sources is merged into one record, keeping the best data from each (for example, OpenAlex citations plus the arXiv PDF). Preprints are matched to their published version by DOI, arXiv ID or title and year.
• Open-access aware. Filter to free-to-read papers and get a direct PDF link where one is openly available.
• Licence-clean. Only openly licensed metadata. No full texts are copied, and abstracts are only included where the licence allows it (arXiv, CC0). Every record carries its licence and source attribution.
• Agent-ready. One query in, predictable JSON out. Works through the Apify API and the Apify MCP server, so Claude, ChatGPT, Cursor and other MCP clients can call it as a tool.
• Fast. A typical search returns in under 15 seconds.

Input

Field	What it does
query	Keywords searched in titles and abstracts (required).
sources	openalex, crossref, arxiv. Default: all three.
yearFrom / yearTo	Publication year range (optional).
openAccessOnly	Only free-to-read papers (default off).
sort	relevance (default), citations (most cited first) or newest.
maxItems	Maximum unique papers to return, 1–500 (default 20).
includeAbstracts	Add abstracts for arXiv papers (default off).
openAlexApiKey	Optional free OpenAlex key for your own daily quota.
contactEmail	Optional email for Crossref's faster "polite" pool.

{
  "query": "retrieval augmented generation",
  "sources": ["openalex", "crossref", "arxiv"],
  "yearFrom": 2022,
  "openAccessOnly": true,
  "sort": "citations",
  "maxItems": 50
}

Advanced: arXiv field syntax such as cat:cs.CL AND ti:retrieval is passed straight to arXiv.

Output

One record per unique paper:

{
  "title": "Attention Is All You Need",
  "authors": ["Ashish Vaswani", "Noam Shazeer", "Niki Parmar", "Jakob Uszkoreit", "Llion Jones", "Aidan N. Gomez", "Łukasz Kaiser", "Illia Polosukhin"],
  "authorCount": 8,
  "year": 2017,
  "publicationDate": "2017-06-12",
  "venue": "arXiv (Cornell University)",
  "publisher": null,
  "type": "preprint",
  "doi": "10.48550/arxiv.1706.03762",
  "doiUrl": "https://doi.org/10.48550/arxiv.1706.03762",
  "arxivId": "1706.03762",
  "openAlexId": "W2626778328",
  "citations": 122093,
  "isOpenAccess": true,
  "oaStatus": "green",
  "pdfUrl": "https://arxiv.org/pdf/1706.03762",
  "landingUrl": "https://arxiv.org/abs/1706.03762",
  "abstract": null,
  "licence": null,
  "source": "OpenAlex",
  "sources": ["OpenAlex", "arXiv"],
  "metadataLicence": "Metadata from OpenAlex (openalex.org), CC0 1.0. Metadata from arXiv.org, CC0 1.0. Thank you to arXiv for use of its open access interoperability.",
  "retrievedAt": "2026-11-03T20:30:00Z",
  "query": "attention is all you need"
}

Field notes:

• citations comes from OpenAlex or Crossref (the higher of the two when both have it). arXiv has no citation counts, so arXiv-only papers show null.
• isOpenAccess is true when OpenAlex flags the paper as open access, Crossref shows a Creative Commons licence, or the paper is on arXiv. null means unknown, not closed.
• licence is the paper's own licence where a source records it (for example cc-by or a Creative Commons URL).
• authors lists up to 50 names; authorCount gives the full number.
• pdfUrl and landingUrl are links only. The Actor never downloads or redistributes full texts.

Pricing

Pay per result: a small fee when the run starts and a fee per unique paper returned. Duplicates merged across sources are counted once. No subscription. Set a maximum cost per run in the run options and the Actor stops when it reaches it.

Data sources and licences

Source	What it adds	Licence / terms
OpenAlex	Citations, open-access status, venues, PDF locations	All data CC0 (pricing & licence)
Crossref REST API	DOI registry metadata, licences	"Almost none of the metadata is subject to copyright, and you may use it for any purpose" (docs)
arXiv API	Preprints, categories, abstracts	Descriptive metadata CC0 (API terms)

Thank you to arXiv for use of its open access interoperability. This Actor is not endorsed by OpenAlex, Crossref or arXiv. It uses each API as documented, identifies itself with a descriptive User-Agent and respects each rate limit (arXiv: one request every three seconds).

Good to know

• OpenAlex quota. Without a key, OpenAlex shares a small free daily quota per network. If it runs out, the run carries on with Crossref and arXiv and says so in the log. Add your own free OpenAlex key in the input to avoid this.
• Open access on Crossref is inferred from a Creative Commons licence, so with openAccessOnly some open papers without a recorded licence are left out.
• Sorting. Crossref and arXiv results are always the most relevant matches, then sorted the way you ask; OpenAlex sorts its title-and-abstract matches directly.
• Up to 500 papers per run. For larger pulls, split by year range.
• Abstracts from OpenAlex and Crossref are not included, because publishers or authors may hold copyright in them.

Use with AI agents (MCP)

Add this Actor to the Apify MCP server (https://mcp.apify.com) and your agent can call it as a tool: pass a query, optionally yearFrom, openAccessOnly and sort, and read back the JSON records. Keep maxItems small (10–20) for fast, cheap agent calls.