Enter your game
Loading the word lists.
Paste share text
Paste the colored squares from a finished game. Green, yellow, white and black squares are read, six to a row. Orange counts as green and blue counts as yellow, for the high contrast palette. Share text has no letters, so you still type the words.
How to review a game
- Type the answer in the first box. Any word on the full six-letter list works, so games from other apps can be replayed here too.
- Type your guesses in order, up to six. A small colored preview appears beside each label as soon as the answer and the guess are complete.
- Press Review my game. One card appears per guess and fills in within a few seconds.
- Read each card from top to bottom: the words left, expected against actual, the best guess at that moment and a one-line verdict.
- If you only have a share grid, open Paste share text, paste it and press Read the colors. To see the sample game below in the tool, open it with its answer and guesses already filled.
A sample game, reviewed
This sequence is illustrative. It is a plausible game with one weak guess, not a record of a real player. The target is FINGER, the daily word for puzzle #41 (April 29, 2026).
Guess 1: MASTER
- Answer list
- 682 to 27 words
- Full list
- 18,195 to 503 words
- Expected and actual
- Expected 9.2, actual 27 (answer list)
- Best guess then
- SINTER, expected 7.2, worst case 26. Top opener on the starter ranking
Fair guess. It was expected to leave 9.2 words on average, while SINTER leaves 7.2 on average. Luck did not help, with 27 left.
Guess 2: COPPER
- Answer list
- 27 to 11 words
- Full list
- 503 to 219 words
- Expected and actual
- Expected 6.3, actual 11 (answer list)
- Best guess then
- WINDOW, expected 1.6, worst case 3
Weak guess. It was expected to leave 6.3 words on average, against 1.6 on average for WINDOW.
Guess 3: DINNER
- Answer list
- 11 to 1 word
- Full list
- 219 to 14 words
- Expected and actual
- Expected 1.0, actual 1 (answer list)
- Best guess then
- DINNER, expected 1.0, worst case 1. It could be the answer
Best guess. No other word was expected to split the possibilities better.
Guess 4: FINGER
- Answer list
- 1 to 1 word
- Full list
- 14 to 1 word
- Expected and actual
- Expected 1.0, actual 1 (answer list)
- Best guess then
- Not needed, the game was won
Solved. It was the only word on the answer list that still fitted.
MASTER cut the answer list from 682 words to 27, a fair opening that left more than expected. COPPER is the weak guess. It was expected to leave 6.3 words and left 11. WINDOW is the stronger probe, expected to leave 1.6. DINNER then took the best remaining guess and FINGER arrived on guess 4.
The full list tells a different story. After COPPER, 219 accepted words still fit, against 11 on the answer list.
Reading each number
Answer list and full list. The first count uses this game’s 682 answer words. The second uses all 18,195 accepted guesses.
Expected. Take every word that still fitted before the guess and treat each as equally likely. Sort them by the colors this guess would give. Expected words left is the size of your group averaged over all of them.
Actual. This is the size of the group that your answer landed in. Over many games the actual numbers average out to the expected one. In a single game they can land far from it.
Best guess then. It is the top word in the ranking over a pool of guesses. The ranking sorts by expected words left first, then by the worst case, then by how many color patterns the guess can produce. A word that can be the answer wins a tie, then alphabetical order decides.
Verdicts. Strong means the guess was within 15 percent of the best expected value, or within half a word. Fair covers up to 60 percent worse or less than two words worse. Weak is anything beyond. A note on luck appears when the actual number was at most half of the expected or at least double it, and the two differ by at least 2 words.
When the answer is on this game’s list, expected and actual count the answer list. Otherwise they count the full list.
What luck and skill look like in the numbers
A reference solver played every one of the 682 answers. It opens with SINTER, the top word of the starter ranking, and then plays the best guess by the same ranking.
After guess 1 the median game had 5 words left and the mean was 7.23. That mean equals the expected value of the opener, 7.23, exactly, because expected is the average of actual over all answers. The typical game sits below the average since a few large groups pull the mean up. The luckiest tenth had 1 word left and the unluckiest tenth had 16 or more, with a worst case of 26. In 13.8 percent of games, more than double the expected number remained.
After guess 2 the median was 1 and after guess 3 it was 1. A game that is solved counts as 1. The solver found the answer by guess 3 in 98.4 percent of games, with a mean of 2.74 guesses and no game past guess 4. A second solver that guesses only words still possible took a mean of 2.76 guesses and needed up to 5.
Skill shows up in the opener you pick. Every displayed word was ranked as an opener against the 682 answers. SINTER ranks first at 7.23 expected words left. OATIER ranks tenth at 7.68 and NITRES ranks 50th at 8.64. The median word leaves 27.12 and ZOOZOO comes last at 350.23.
Put side by side, the two effects differ in size. Choosing the tenth best opener instead of the best costs 0.45 words on average, and the median opener costs 19.9. One game, by contrast, moves by 15 words between the lucky and unlucky tenths. Among the top openers luck settles the first turn, while an average opener gives up more than luck usually does.
Share text and other games
Paste share text reads rows of six squares. Green and orange squares count as green. Yellow and blue squares count as yellow. White and black squares count as gray. Lines without squares, such as a title, are skipped. A row with the wrong number of squares or a grid longer than six rows is rejected with a message that says why.
If you type the answer, its colors win and the page warns when a row disagrees with the paste. Without an answer the review runs from the pasted colors alone.
To solve a live board instead, use the 6 Letter Wordle Helper.
How the counts are made and where they stop
The counts come from replaying each guess against the word lists with the game’s own feedback rule. The answer list is this game’s own list and not an official Wordle list. The full list of 18,195 words descends from the Wordnik word list, which carries an MIT license, which the About page credits in full. Best guesses are drawn only from the 18,094 displayed words, so 101 words that could offend are never suggested, though they still count as possible answers.
To keep phones responsive, the ranking stops at 400,000 comparisons. Sets of up to 22 candidates are checked against every displayed word. Sets of up to 586 are checked against the answer list alone, and sets of up to 632 against themselves. A larger set stays unranked, except on an empty board, where the starter ranking’s top word stands in.
The review treats every remaining word as equally likely and knows nothing about what you had in mind. It measures how well a guess splits the possibilities.
Frequently asked questions
Do I need a real game to use the review?
No. Type any answer and any guesses from the full list. That makes the review useful for practice, for a game you remember from a friend or for testing how a different line of play would have gone.
What does expected words left mean?
It is the average number of words that would remain after the guess, taking every word that was still possible as equally likely to be the answer. A lower number means the guess splits the possibilities into smaller groups.
Why does the best guess sometimes look like an odd word?
The ranking looks for the guess that separates the remaining words best, and that word does not have to be a likely answer. When a few candidates remain, an unusual word with the right letters can tell them apart in one turn.
Can I review a game from another app?
Yes, as long as every word is on the full list of six-letter words. When the answer is not on this game's answer list, the review counts against the full list instead and says so on each card.
Why can't the review work from share text alone?
A share grid holds colors but no letters. The review needs the words to count what each guess left and to compare it with other guesses. You can paste the grid to fill the colors, then type the words.
Is my game saved or sent anywhere?
No. The words you type stay on your device. The page stores nothing, its code sends no request that carries them and the address bar is never changed. If you reload, the boxes start empty unless the link you opened already contains a game.