Nutristatic Indexes

These are the prebuilt phrase-frequency indexes. Each lives at a permanent versioned URL (/idx/<year-month>/…, the dump it was built from), so a link that names an index keeps meaning the same data even after newer editions appear. Current edition: 2026-08. Download one to search offline: open the offline version, then pick the .index file you downloaded. You can also point the web app at any of them with a custom URL, or host them yourself.

Available indexes

Each index also has a compressed .idxz sidecar (roughly half the size) used for range-mode self-hosting — append .idxz to any download link. You do not need it for offline search; the .index file is enough.

Sources & licensing

• Wikipedia indexes — built from the August 2026 Wikipedia database dumps. The article text is written by Wikipedia's contributors and is available under the Creative Commons Attribution-ShareAlike 4.0 License (CC BY-SA 4.0). These indexes are word- and phrase-frequency data derived from that text, and are shared under the same license. Attribution: Wikipedia contributors; dumps from dumps.wikimedia.org. Diacritics are folded to their base letters (and German umlauts to ae/oe/ue/ss); the search box applies the same folding to whatever you type.
• web words + bigrams (the demo index) — built from Peter Norvig's ngram frequency lists, themselves derived from the Google Web Trillion Word Corpus. Single words and two-word phrases only.
• Index format & search engine — a serverless rewrite of Dan Egnor's Nutrimatic; the index format is byte-compatible with its tools. The Nutristatic code is open source under the GPL-2.0 license.

« Back to search

Impressum · Datenschutz