Wordle Lab
Wordle Lab – Ultimate Wordle Solver & Analyzer

Here are three commonly referenced word lists associated with the game Wordle: the “2,309”, “3,207”, and “14,855” word lists. None of these lists have been officially confirmed as the definitive answer sets by the creators of Wordle.

• 2,309 Word List:

  Many players report seeing answers from this 2,309‑word list, though there are exceptions.

• 3,207 Word List:

  The 3,207‑word list is often used as a broader “answers” list in community tools.

• 14,855 Word List:

  This much larger list, containing 14,855 words, is believed to reflect the broader vocabulary of words accepted by the game as guesses (rather than necessarily as answers).

Each of these lists plays a different role: the 2,309 list as a common answers reference, the 3,207 list for extended answer coverage, and the 14,855 list for comprehensive guess vocabulary.

• Repeated letters (important): Basic Mode does not enforce exact repeated-letter counts. Example: If the answer is EJECT, a guess can produce both 🟩E and 🟨E, which implies at least two E's. Basic Mode won’t encode that “at least 2” constraint precisely. Use Advanced Mode → Pattern Search for exact filtering (e.g., E*E).

• E[steps]: The expected number of steps to solve from the current state under a specified guessing policy. The smaller, the better.
  • Note: the policy used for E[steps] may differ across panels (see panel-specific notes).
• Entropy: A measure of uncertainty. Higher values indicate guesses that are more informative. The bigger, the better.
• E[cands]: The expected number of candidates remaining after a guess. The smaller, the better.
• Max Bucket: The size of the largest bucket induced by a guess. The smaller, the better.

• First Guess Advisor shows exact precomputed values for the first guess under the selected Answer Word List and Mode.
• First Guess Advisor requires the Word List at the top of the page to be set to 2,309 or 3,207.
• Answer Word List refers to the Word List selected at the top of the page (e.g., 2,309 or 3,207).
• E[steps] here is exact (not simulated) and is computed under the same mode’s guessing policy:
  • Hard Mode: all guesses are restricted to the remaining candidates.
  • Non-hard Mode: follow-up guesses are allowed from the selected guess pool (Answer Word List or 14,855), consistent with the mode.
  • In non-hard mode, this differs from the Next Guess Recommender and Word Analyzer, which still restrict follow-up guesses to remaining candidates when computing E[steps].
• Therefore, E[steps] here may differ from the Next Guess Recommender and Word Analyzer E[steps].

• Next Guess Recommender uses an entropy-based approach to rank candidate guesses from the current state.
• Answer Word List refers to the Word List selected at the top of the page.
• E[steps] is an approximation of the expected number of steps from the current state.
  • All follow-up guesses used for E[steps] are chosen only from the remaining candidates at each step, regardless of whether Hard Mode or Non-hard Mode is selected
  • Hard Mode affects where recommended guesses come from:
    • Hard Mode: guesses are selected only from the remaining candidates.
    • Non-hard Mode: guesses are selected from both the remaining candidates and an external guess pool, based on entropy.
  • Max Cand (Max Candidates): the number of top-ranked candidates (by entropy) considered as possible guesses.
  • Max Pool (Max Guess Pool): the number of top-ranked external-pool words (by entropy) considered as possible guesses (only in Non-hard Mode).
  • Deep Limit: controls when the recommender switches between the fast and deeper E[steps] models:
    • If the number of remaining candidates is greater than this value, the recommender uses a faster, shallower E[steps] approximation.
    • If the number of remaining candidates is less than or equal to this value, it switches to a deeper multi-step lookahead, which improves accuracy but can be significantly slower.
• For the very first guess, use First Guess Advisor above for exact precomputed rankings.

• Word Analyzer: Lets you type any 5-letter word and evaluate it against the current candidate set. It computes the same metrics as the Next Guess Recommender: Entropy, E[steps], E[cands], and Max Bucket, but for a single word you choose.
• E[steps]: An approximate expected number of guesses needed to find the answer if you start by using this word, given the current candidate pool.
  • Just like in the Next Guess Recommender, follow-up guesses used for E[steps] are always restricted to the remaining candidates at each step, not from the external pool.
  • Deep E[steps] (checkbox): When enabled, the Word Analyzer uses a much deeper lookahead model for computing E[steps], offering a result closer to the true expected number of steps. This mode is considerably slower on large candidate sets. When unchecked, the faster shallow approximation is used instead.
• Pattern Distribution: For the entered word, the analyzer also shows a full breakdown of all feedback patterns (🟩🟨⬛ combinations) and how many candidates fall into each pattern. This view helps you understand:
  • how the word partitions the remaining candidates,
  • how balanced the resulting branches are,
  • why entropy, E[cands], and Max Bucket have their specific values.

• ?: Exactly one letter (e.g.,A?E→ACE, APE, ARE)
• *: Zero or more letters (e.g.,E*E→AGREE, EAGLE, EERIE)
• [AEIOU]: Any one listed; [^AEIOU]: Not these (e.g.,[^AEIOU]ATE→SLATE, CRATE)
• &: AND; |: OR (AND has higher precedence)
• ( ): Group terms (e.g.,(ING & R) | (A?E & N))

This tool is an independent, unofficial helper for the game Wordle. It is provided for educational and entertainment purposes only.

All word lists, recommendations, and computed results are provided “as is” with no guarantees of accuracy or completeness. Actual Wordle answers may differ from the words shown or suggested here.

Wordle is a trademark of The New York Times Company. This site is not affiliated with, endorsed by, or associated with The New York Times or Wordle.