Wordle Solver

Type your five-letter guess into the grid. Letters start gray. Click or tap a tile to cycle gray → yellow → green; on a desktop keyboard, the 1, 2, and 3 keys select those states directly.

Copy the feedback exactly, then press Solve. Use “Add guess row” for another round. All rows are evaluated together, including the position of every yellow tile and the exact bounds implied by duplicate gray, yellow, or green letters.

The list is ranked by coverage within the remaining candidate pool. This is a fast, transparent heuristic rather than a full entropy simulation, and it does not guarantee that the first result is the answer used by a specific Wordle edition.

An opening word with five distinct letters checks more letter identities than an opening with a duplicate. Balanced examples such as CRANE or SLATE combine vowels and consonants, while a vowel-heavy word such as AUDIO answers a narrower question about the vowel pattern. No single opener is best for every answer list.

After an all-gray result, a second guess made from unused letters can rule out more of the alphabet. Once you have yellow or green feedback, decide whether you want a valid remaining candidate or an information-gathering guess; hard-mode rules may require every revealed letter to be reused.

A yellow letter must appear again in a different position, while a green letter stays fixed. Duplicate letters need extra care: one gray copy does not always eliminate that letter if another copy in the same guess is yellow or green.

Each guess divides the possible answers into groups defined by their color patterns. A useful information-first guess tends to avoid leaving most answers in one large group, but the exact choice depends on the answer and accepted-guess lists used by that Wordle edition.

A complete optimization would simulate every possible feedback pattern and compare the sizes of the groups it creates. Results from one published answer list cannot be assumed to transfer unchanged to another list, so Kelimator does not label one universal starting word as mathematically best.

Our solver does not compute a full decision tree in real time. It combines distinct-letter frequency with position frequency across the remaining candidates. That makes the order reproducible and useful for exploration without presenting it as a mathematically optimal strategy.

The most common input mistake is leaving a yellow tile gray. A yellow letter is in the answer but not in that position, so cycle the corresponding grid tile to yellow.

The second common mistake is with duplicate letters. If a guess contains two E tiles and one is green while the other is gray, the answer contains exactly one E. The solver handles this automatically when each tile in the grid has the same color shown by the game.

Third is using the solver before your first guess. With no constraints, the solver returns thousands of words — that's not actually useful. Pick a good starter yourself; the solver shines on turns 2–4.