API reference

The OpenAlex API is a fast, modern REST API over the whole OpenAlex dataset — a connected graph of works, authors, sources, institutions, topics, and more. You can try it right now, no account needed: paste a URL into your browser and you get JSON back.

Every entity type is an endpoint (/works, /authors, /sources, …), and you work with each one the same handful of ways: list, filter, search, sort, group into counts, page, and select just the fields you want — or fetch a single entity by its ID. What each entity and field actually means lives in the Data reference; these pages are the wire mechanics.

Basic use is free. A free API key raises your daily budget 10×, and heavier use is pay-as-you-go. New here? The Quickstart walks you through your first real query.

Good to know

A few things that surprise people early:

• Filter by IDs, not names. Names are ambiguous — “Smith” is thousands of authors, “MIT” several institutions. Look up the entity, grab its OpenAlex ID, and filter on that. Already holding a DOI, ORCID, ROR, or ISSN? Use it directly and skip the lookup.
• Nested entities come back dehydrated — a trimmed stub with just an ID and a display name. Fetch the full record separately when you need the rest.
• By default you’re querying the curated core corpus of 300M+ works. A much larger expansion (mostly datasets and repository records, formerly called “XPAC”) is opt-in via the corpus parameter, so don’t be startled when corpus=all adds roughly 60% more works to your counts.
• Everything is snake_case, and all data is CC0 — no license worries, ever.
• Pointing an AI agent at the API? Hand it the LLM Quick Reference; there’s a machine-readable OpenAPI spec too.

For the complete endpoint list see the Endpoints index; for expressive, saveable queries, see OQL.

Making a request

The base URL is:

https://api.openalex.org

Send your key as an api_key query parameter (or leave it off to try the API for free — the website itself runs on these same public endpoints):

curl "https://api.openalex.org/works?filter=publication_year:2024,is_oa:true&per_page=10&api_key=YOUR_KEY"

That’s a list-and-filter request: open-access works from 2024, ten per page. See Authentication for keys and rate limits, and the Querying pages for the full filter, search, sort, group, and paging syntax.

Response format

Every list endpoint returns the same envelope:

{
  "meta": {
    "count": 286750097,
    "page": 1,
    "per_page": 25,
    "cost_usd": 0.0001
  },
  "results": [
    { "id": "https://openalex.org/W2741809807", "title": "...", ... }
  ],
  "group_by": []
}

Security

Text fields in OpenAlex (and Unpaywall) records — titles, abstracts, affiliation strings — come from external sources, and we pass them through as-is rather than sanitizing them. If you display or process this text, especially on web pages, follow standard best practice for handling untrusted text: escape it for your output format, or use tools that do it for you (HTML templating engines, prepared statements). We’ve never seen a malicious payload in OpenAlex or Unpaywall data, but it’s possible in principle, and following best practice is cheap. Good starting points: OWASP’s XSS prevention cheat sheet and Wikipedia’s overview of code injection.

This API reference is split into four sections: