Editorial Policy and Methodology

Effective September 3, 2026

Primary data

Puzzle dates, numbers, answers, and Connections boards come from New York Times puzzle endpoints. Responses must match the expected structure before they can be published.

What is computed

Vowel count, first and last letters, second letter, repeated-letter status, display patterns, and progressive candidate counts are calculated locally. The 0–100 difficulty score is a percentile calibrated across 2,309 possible answers using positional letter rarity, with disclosed adjustments for repetitions and words without standard vowels. It is not an official NYT rating or a prediction of individual solve time.

Use of AI

A generative model may draft introductions, definition-style clues, and commentary. Its output is schema-validated. Deterministic letter facts overwrite generated values, and a dictionary definition is preferred when available. A non-AI fallback is published when generation fails.

Solver ranking

The solver keeps a 2,309-word possible-answer list separate from a 12,972-word allowed-guess list. It reproduces duplicate-letter feedback, partitions every remaining answer by the feedback each evaluated guess would produce, and reports entropy, expected remaining candidates, and the worst-case bucket. It evaluates the full allowed list at 250 or fewer candidates and an explicitly labeled 400-word shortlist on larger sets to keep browsers responsive. It does not promise a globally optimal multi-turn strategy.

Corrections

If a clue, answer, attribution, or technical behavior appears wrong, please send a correction. Material factual errors are corrected as soon as they are confirmed.

Independence

Wordle Hint Today is independently maintained by Aneeza and is not affiliated with or endorsed by The New York Times.