FreeNIC
Free, unified US banking regulatory data.
FreeNIC v1.1.0 is here
The first stable release harmonizes 4.97B observations across 58 base tables and 20 public datasets — 217,210 institutions, 1863–2026 — into one queryable schema, published as a 67-file, 13.9 GiB parquet release at data.freenic.org. New in v1.1.0: a two-explorer browse surface — institution lookup over the 59,824 active institutions and a bank-failure explorer over the 4,115 FDIC failure records. Everything shipped is real published data; nothing is fabricated. See what's in the release.
Explore the data
Interactive charts over real, published FreeNIC tables — bank failures, FRED banking series and active institutions — each with a top-right Download CSV. Browse and chart in your browser, no install.
The browsable explorer shows a curated sample (15.5 MB) of the FreeNIC database; the complete dataset (4.97B rows, 58 base tables, 1863–2026) is available via the R/Python package and API over the parquet release at data.freenic.org.
Data dictionary
The harmonized MDRM/UBPR schedule line-items — the version-pinned bank-data-dictionary taxonomy mapping codes to schedules, captions and edit relationships. Download the core table, or browse and search the full dictionary →
Four ways in
Pick the smallest one that answers your question.
Explorer →
Browse + chart the curated sample in the browser. curated sample
Data download →
The full v1.1.0 parquet release — 67 files, 13.9 GiB — and per-table CSV/Parquet. live
Python / R / API →
The freenic_py / freenic_r packages + MCP server;
remote DuckDB httpfs over the parquet release.
read layers
GitHub →
The full ingestion pipeline and read packages — MIT code, reproducible from public sources.
More on what FreeNIC is, how to query the full release remotely, and the public-domain (CC0) historical layer → Methodology & provenance.