Written by Pop. Dataset documentation, first published October 28, 2026, compiled from the level guides on this site and two public word lists.
This page documents the first open dataset describing every Wordscapes level as numbers: 6,366 rows covering the 6,000 main levels and the 366 fixed daily calendar slots. Each row holds the word counts, the word length distribution, the bonus and refused word counts, the rarity of the board words on the Zipf frequency scale, and a documented difficulty score from 0 to 100. The file is free to download and free to reuse under a CC BY 4.0 license.
What the dataset will not give you
No circle letters, no board words, no bonus words, no solutions. The file cannot solve a single level: it measures the levels without reproducing them. For the actual answers, every row carries a link to that level’s guide on this site, and the 366 daily rows point to the daily puzzle answers hub where today’s board lives.
Five things the numbers show
- The game front-loads its difficulty. Levels 1 to 1,000 average 55.0 on the 0 to 100 difficulty scale, while levels 5,001 to 6,000 average 47.5. The scores drop band after band, so the climb most players expect never happens.
- The hardest board sits early. The top score of all 6,000 main levels belongs to Wordscapes level 1431 with 80.9: 18 board words, 15 bonus words and 33 valid words the level refuses. Eight of the ten hardest boards sit below level 1500.
- Refused words are the rule, not the exception. 99.1 percent of main levels turn down at least one valid English word from the public ENABLE list, a behavior our study of the words Wordscapes refuses measured across the whole game.
- Rare words are everywhere. 82.3 percent of main levels contain at least one board word with a Zipf frequency below 3.0, which is rarer than roughly one occurrence per million words of English.
- Daily boards are built differently. The 366 daily slots follow their own difficulty curve, and four special slots use larger letter circles of 8 or 9 letters.
What is inside
| Column | Meaning |
|---|---|
level_id, level_type |
1 to 6000 for main levels, daily-001 to daily-366 for the daily calendar slots |
letters_count |
how many letters the level’s circle holds |
words_total, words_len3 to words_len9 |
how many words the board holds, in total and per word length |
longest_word_len, avg_word_len, allowed_lengths |
the shape of the board’s words and the word lengths the level accepts |
bonus_words_count |
how many bonus words the level accepts |
rejected_valid_words_count |
how many valid ENABLE words the level refuses |
avg_word_zipf, min_word_zipf, rare_words_count |
how common the board words are on the Zipf frequency scale |
difficulty_score, difficulty_percentile |
a documented 0 to 100 score and its percentile |
source_url |
the published guide for that level |
How the difficulty score works
The score blends four components, each computable from the columns of the file itself: the rarity of the board words (weight 0.35), the number of board words (0.25), their average length (0.15), and the share of valid swipes that do not advance the board, meaning bonus and refused words (0.25). Each component is scaled on the 6,000 main levels and the weighted sum is mapped to 0 to 100. The exact formula, bounds and weights are written out in the README that ships with the file, so anyone can reproduce or challenge the score.
Method in one paragraph. The figures are compiled from the level guides published on this site, the public domain ENABLE word list and the wordfreq frequency data. A word only counts as refused on a level whose allowed word lengths include its length, because levels skip some lengths by design. Levels that changed board over time are counted once, on their current board. Under these rules the 6,000 main levels yield 15,612 distinct playable ENABLE words, the same figure as the published study.
Download and license
The dataset is released under the Creative Commons Attribution 4.0 license: free to use, share and adapt, including commercially, with attribution to Pop, SolvingWordscapes.com and a link to this page.
- CSV download:
wordscapes-level-metrics-v1.csv(6,366 rows, 22 columns) - Zenodo archive: the permanent record sits at DOI 10.5281/zenodo.23737776, with versioning for the future master levels update
- Kaggle mirror: the dataset on Kaggle is the home for notebooks and community analysis
- Hugging Face mirror: the dataset on Hugging Face includes a built-in table viewer for all 6,366 rows
How to cite
FAQ
Can the answers be rebuilt from this file? No. The file contains counts, frequency statistics and links, never letters or words. Knowing that a board holds 4 three letter words and 3 four letter words says nothing about which words they are.
Why ENABLE and not the game’s own word list? ENABLE is public domain and widely used as a baseline in word game research. The gap between what ENABLE considers valid and what the game accepts is precisely what the refused words column measures.
Will master levels be added? Yes, a version 2 covering the master levels is planned. Version numbers and changes are tracked in the README and on the Zenodo record.
Can I use it in a notebook, a paper or an article? Yes, that is what it is for. CC BY 4.0 only asks for attribution with a link.
SolvingWordscapes.com is an independent fan site and is not affiliated with PeopleFun or Wordscapes.