Open data
The whole catalogue (6,319 courses, 6,922 offerings) is free to download and rebuilt with every update. The large files are gzip-compressed. Last built Nov 9, 2026.
- courses.ndjson.gz
Every course, one JSON object per line, with all of its offerings. The most complete format.
- offerings.csv.gz
One row per way to take a course: a flat table for spreadsheets.
- institutions.json
Institutions with ROR ids, countries and course counts.
- schema.json
JSON Schema for a course record in courses.ndjson.
What's in each record
A course (e.g. MIT 6.006) has one or more offerings: concrete ways to take it, like an archived course site, a lecture playlist or a scheduled run. Each offering has an outbound URL, an access label (free, free-account or free-paid-extras), pacing and dates. Subjects are UNESCO ISCED-F 2013 codes; languages are BCP-47 tags; institutions carry ROR ids where we have them. The full format is described by schema.json.
Licence
Facts (titles, links, dates, subjects, identifiers) and the structure we add are free to reuse. Course descriptions are written by the institutions and keep their original licences, so check the source before republishing them. Each record lists where it came from.
Updates
The files are rebuilt after every harvest, usually daily. Every version is kept, so if you need an earlier snapshot or a different format, get in touch.