fshy

Home / Debrid

OpenAIRE Graph API

graph.openaire.eu/docs/Scholarly graph aggregator (publications, datasets, software, projects)debrid/openaireJSONcomparereport an errorprovider JSON

Free allowance
Public REST APIs over 'one of the largest open scholarly record collections worldwide', with no registration described on the documentation landing page.
Verdict
★ qualifiesusable as stated; the read path was probed
Cost
free tier
Identity
anonymousno card, no phone
Binding limit
request-ratenone published on the docs landing page; two calls here (size=1 and size=100) both returned 200 with no throttling header
Evidence
measured3 sources; a live request was made and its output kept
Freshness
stale on 2026-11-23checked 2026-10-23 (live); stale on 2026-11-23
Free tierPublic REST APIs over 'one of the largest open scholarly record collections worldwide', with no registration described on the documentation landing page. The Graph itself is 'openly available and can be used by third-parties to create added value services'.
unit unlimited; metered on none; No key and no numeric rate limit appear on the docs landing page.
PriceUSD 0
Limits
  • request-rate (binding) none published on the docs landing page; two calls here (size=1 and size=100) both returned 200 with no throttling header source
Bindingrequest-rate At 8.38 MB per call and 4.2 s per call, one connection moves about 2 MB/s; with no published cap the unknown is how many calls the service tolerates before a 429, which is what decides the day.
Identityanonymous No key, no account: both calls answered 200 anonymously. Nothing on the docs page asks for registration.
EgressFree and unmetered for the public API endpoints.
free allowance: no byte allowance published
Per-object capnot applicable: XML or JSON per query. size=100 returned 8,384,696 bytes in one response here, so a page is megabytes rather than kilobytes.
Retentionthe graph is rebuilt from source repositories and deduplicated; records reflect the last harvest, not a fixed snapshot
Interfacerest, cli GET https://api.openaire.eu/search/publications?title=...&size=N (XML by default, format=json available); the newer Graph API sits under https://api.openaire.eu/graph/.
Toolingcurl/python-requests; the openaire python client exists but the endpoints are plain HTTP.
Composes withThe widest single metadata source per call: 8.4 MB per size=100 page means a few thousand calls moves tens of GB of linked records, which is a different order from Crossref's 0.77 MB pages.
ReliabilityResponses are large XML by default, so a client that forgets format=json spends bandwidth on tags; the search endpoint and the newer graph endpoint are different APIs with different shapes.
NotesThe highest bytes-per-call of any metadata API by an order of magnitude, which makes it the best candidate for filling a day with metadata when the file hosts are the bottleneck.
Evidence notemeasured Two calls were made and their byte counts recorded, one of them 8.4 MB. The absence of a rate limit is an absence of evidence: no sustained run was attempted.
Parsed numbers
  • has-free-tier 1
  • price USD 0
Evidence 3
datesourcewhat it proved
2026-10-23graph.openaire.eu/docs/'one of the largest open scholarly record collections worldwide' and that the graph data are 'openly available' through public APIs
2026-10-23probe: https://api.openaire.eu/search/publications?title=cancer&size=1200, 24,711 bytes of application/xml, anonymous
2026-10-23probe: https://api.openaire.eu/search/publications?title=cancer&size=100200, 8,384,696 bytes in 4,166.8 ms - one anonymous call returns 8.4 MB
More fields 3
size 1 bytes24711
size 100 bytes8384696
size 100 ms4166.8