Crossref REST API

crossref.org · Debrid · 1 record · JSON

Crossref REST API Debrid

★ qualifiesanonymousno cardDrequest-rate

www.crossref.org/documentation/retrieve-metadata/rest-api/ · scholarly metadata index (DOI registry) · debrid/crossref-rest-api · JSON · compare

Free tier	not published unit unlimited; metered on none; No numeric rate limit is published. The documented guidance is 'back off if you start seeing 429 statuses'.
Price	USD 0
Per-object cap	not applicable: paged JSON. rows=1000 with a select list returned 773,304 bytes in one response here.
Retention	the registry only grows; records are corrected rather than deleted, and the API is the live view
Egress	Free and unmetered. No key, no account, no allowance published. free_allowance: no byte allowance published
Identity	anonymous No sign-up is required. A mailto in the query or User-Agent is a courtesy that routes the caller to the polite pool; it is not a credential.
Limits	request-rate no number published; the tips page says 'pay attention to the http status code and back off if you start seeing 429 statuses' source per-object-cap cursor paging is required past the first page; rows=1000 is accepted and returned 773,304 bytes here The documented bulk snapshot did not work: https://api.crossref.org/snapshots/monthly/latest.json.gz answered 200 with 41 bytes of text/html and no body, and /snapshots answered 200 with a zero-byte body. That is a false-200 shape, not a file.
Binding	request-rate At 773 KB per rows=1000 call, 1 TiB is about 1.4 million calls; the call rate, which is unpublished and signalled only by 429, is what decides the day's bytes.
Interface	rest, cli GET https://api.crossref.org/works with rows, cursor and select; content negotiation for single records. The polite pool is signalled with mailto= or a User-Agent carrying an address. endpoint: https://api.crossref.org/works
Tooling	curl/python-requests; the harvester package and the documented cursor pattern. No SDK required.
Composes with	Crossref supplies the DOI-level citation graph and licence metadata; the full text behind each DOI is a separate fetch from the publisher or from an OA mirror such as PMC or Europe PMC.
Reliability	The polite pool is the difference between a fast run and a 429 storm, and it costs only an email in the User-Agent; the snapshot endpoint is silently broken and returns a 200 stub, which a naive client will save as a data file.
Notes	Metadata-only: the corpus is descriptions of papers, not papers.
Evidence rung	documented Two endpoints were exercised with byte counts and the vendor page was read. No 429 was triggered here, so the unpublished rate ceiling is not characterised.
Checked	2026-10-24 (live); stale on 2026-11-23
Parsed numbers	`$.price_per_tb_month` price-per-tb USD0.0 TB / month

Evidence (4)

date	source	what it proved
2026-10-23	www.crossref.org/documentation/retrieve-metadata/rest-api/	'No sign-up is required to use the REST API' and the polite-pool mailto pattern
2026-10-23	www.crossref.org/documentation/retrieve-metadata/rest-api/ti	'back off if you start seeing 429 statuses' - the only published rate signal; cursor paging for bulk
2026-10-23	probe:https://api.crossref.org/works?rows=1000&cursor=*&select=DOI,title,container-title,published,author	200, 773,304 bytes of application/json in 3,456 ms - the real bytes per call
2026-10-23	probe:https://api.crossref.org/snapshots/monthly/latest.json.gz	200 with 41 bytes of text/html and no file - the documented snapshot URL did not serve a snapshot

More fields (2)

{
 "rows_1000_bytes": 773304,
 "snapshot_status": "200 with 41-byte HTML stub"
}

Unknown or withheld values (1)

$.free_tier.stated withheld