25,798 datasets from 394 official sources: IMF, World Bank, Eurostat, FRED, national statistics offices. Searchable and queryable on your machine.
Free forever · no account · works offline. Then subsets sync, a working copy in under a minute.
Prefer not to run anything? Get a hosted key: 500 free queries, no card →
From pip install to a cited number in under a minute. Every result carries its source, license, and last-updated date.
$ subsets sync ✓ small bundle synced; your copy, works offline $ subsets search "malaysia inflation" 1. dosm-cpi-headline-inflation CPI Headline Inflation · Dept. of Statistics Malaysia 2. world-bank-inflation-consumer-prices Inflation, consumer prices (annual %) · World Bank $ subsets query 'SELECT * FROM "dosm-cpi-headline-inflation" LIMIT 3' ┌────────────┬───────────────┐ │ date │ inflation_yoy │ ├────────────┼───────────────┤ │ 2026-04 │ 1.4 │ │ 2026-05 │ 1.2 │ │ 2026-06 │ 1.1 │ └────────────┴───────────────┘ source: Department of Statistics Malaysia · CC BY 4.0 · updated 3h ago
subsets syncThe small bundle · ~3,500 tables · ~200 MB · under a minute
subsets sync --bundle mediumThe mid-size tier · more sources, more history
subsets sync --bundle allThe full catalog · 138.2 GB
You sync a bundle, small, medium, or everything, and it’s yours. Re-sync pulls only what changed. Local queries answer in milliseconds: on a plane, in CI, forever.
Full catalog, full search, no account. The data is open, the pipelines are MIT; you can walk away with everything.
pip install subsetsio
Same catalog and search as an API / MCP endpoint we run. Nothing to install, always fresh.
Sign up free: instant key, no card500 free queries, then from $19/mo
If you clone the library and never pay, that’s working as intended.