A lightweight, low-cost REST API over the full Crossref corpus

Every work in Crossref's public data file, answering the same requests as api.crossref.org. Runs on one machine: 6 cores, 64 GB of memory, 1 TB of disk.

Open the console API docs

Capacity, one server

Throughput

Capacity by request type

Load-test rates per second, multiplied out to a 30-day month. CPU or 1 Gbit port, whichever limits first; zstd responses. Log scale.

Request	Req/s	p50	p99	CPU each	Size	zstd

Mixed traffic

Requests a second, by connections.

CachedCaches off, 16 connections

Mixed latency

Milliseconds, by connections.

p99p90

Cost

Server rent.

How we tested scale

Load generated from random real requests, on the server and from the internet.

From the internet, through Cloudflare

Test	Connections	Req/s	Round trip p50	p99	In server p50	Server CPU	Errors

Accuracy

Against api.crossref.org, works created before 2026-03-01.

Compatibility

Same paths, parameters, and responses.

Routes

/works/{doi}	Served
/works	Served
/members/{id}/works, /prefixes, /funders, /journals, /types	Served
filter, facet, sort, order, rows, offset, cursor, sample, select	Served
query, query.bibliographic, query.author…	Not yet
/members, /journals, /funders metadata	Not yet

Facets and sorts

affiliationsort: citation

Filters

Use it

No key. 50 requests a second per visitor.

Console

Build and send requests, read results, run a load test from the browser.

Docs

Routes, parameters, filters, facets, cursors, errors, and limits.