What each kind of source gives you

This table is about kinds of source, not products. The row for numi is the one you can check yourself with a sandbox call. The other rows summarise what the research found in the sources it read on 2026-10-29, and a product of a kind can differ from its row.

Kind of source	Keeps the axis	Labels the rows	States a reconciliation to the total	History depth	Licence at the entry tier	Price band
The SEC’s entity-wide JSON APIs	No segment facts: only facts that apply to the whole entity	Not applicable	No	varies	Public information	Free
The SEC’s bulk data sets	Yes in the notes set, as `axis=member` strings with the namespace prefixes removed	No	No	Face-statement set from January 2009; varies for the notes set	Public information	Free
XBRL-to-JSON converters	Yes, as raw facts of one filing	No: member names as filed	No: the caller sums	varies	varies	From tens of dollars a month
Labelled segment extracts	No: the axis is dropped or folded in the products read	Yes	Not documented as a field	varies	Personal use at most entry tiers read	From tens of dollars a month
Hand-built institutional datasets	No in the products read	Yes	varies	varies	varies	Quote
numi	Yes: `axis_qname` and `member` as filed	Yes, made from the member name	Yes: `coverage_ratio` and `reconciliation` for each axis	each filing’s own periods; no depth stated	Commercial use, no attribution, on every plan	Free sandbox; from tens of dollars a month for Builder

Getting started

Data

Guides

API reference

Changelog

What each kind of source gives you