Wordle Answer by Puzzle Number

How Wordle Puzzle Numbers Work

The number in the corner of the Wordle board is a day counter. It started at 1 on 19 June 2021, the day Josh Wardle published the first puzzle, and it has advanced by exactly one every calendar day since. There are no skipped numbers, no resets, and no pauses. The number you see on your screen tells you how many daily puzzles had been published up to and including the one you are looking at.

That makes the whole thing arithmetic rather than a lookup table. Take the date of the puzzle, count the days back to 19 June 2021, and add one. Puzzle 100 was 26 September 2021. Puzzle 500 was 31 October 2022. Puzzle 1,000 was 14 March 2024. Puzzle 1,500 was 27 July 2025. Every one of those is a plain date difference, and you can check any of them against the number in the game's header without trusting anybody.

Where the confusion usually comes from is the word "day". A lot of trackers and a lot of social media call 19 June 2021 "day 1", so the same puzzle gets called day 1 in one place and puzzle 1 in another. Both labels are used, and both are off by nothing and something depending on who is counting. The game itself uses the count-from-launch convention, and so does this page. When you see a number in the Wordle interface, that number is the puzzle number. If a blog post says "day 640" and the header says 640, they agree. If a spreadsheet starts at zero, it will read one lower than the header, which is the other common source of arguments.

There is one small wrinkle worth knowing about, because it explains most "the answer was different for me" reports. The puzzle flips at midnight in your own time zone, not at a single global moment. A player in Sydney can be a full day ahead of a player in Los Angeles, and both of them are correctly on "today's" puzzle at the moment they play. Spoiler posts usually quote the US Eastern date, because that is where the publisher is based, which means a word shared during a New York afternoon is often tomorrow's word for readers in Asia and Australia. The puzzle number travels with the date, so the number is always the reliable half of the pair: if your header says 1,690, you are on puzzle 1,690 no matter which side of the planet you typed it.

The practical upshot is that a puzzle number is a permanent address. A date can be ambiguous across time zones, and it changes meaning if you are looking at a repeat. The number does not. That is why it is worth keeping a list of them, and why this page exists.

Milestones in the Puzzle Number Line

The counter gives you a spine to hang the game's history on. These are the checkpoints below, and every one of them can be clicked in the strip near the top of this page.

Puzzle 1,000 is worth a second look, because the round number is not a round date. It landed on a Thursday in mid-March 2024, which means the thousandth puzzle was an ordinary weekday puzzle with no ceremony attached. That is the normal shape of this counter: it is a plain day count, so round numbers arrive on whatever weekday the arithmetic lands on rather than on a memorable occasion. It is also why the number is such a good archival handle. Nobody has to remember when something happened; they only have to remember a digit and a half.

Puzzle 1,690 is the other one to know, because it is the point where the rules of the archive changed. It is discussed in full below, but the short version is that it is the first puzzle whose answer had already been used. Before it, every puzzle for four years and seven months had produced a word nobody had seen before.

Why People Look Up an Answer by Number

Most people arrive at a puzzle number because they are trying to settle something, and the "something" is usually one of four things.

The first is streaks. Every Wordle client keeps a current streak and a longest streak, and both are expressed in days, but the day that broke a streak is only identified by its number. When someone on a 400-day run misses a day and the streak resets, the useful question is not "what day was that" but "what was puzzle N". If you can name the puzzle, you can pull it up and look at the board, see whether it was a genuinely hard board, and decide whether the miss was skill or bad luck. The answer for a given day is the same data seen from the other direction.

The second is the daily journal. A lot of players keep a written record, and the natural key for a written record is a number, because a number is shorter than a date and does not need a month and a day. Writers who log a puzzle's answer, their guess count, and their starting word use the puzzle number as the entry key. When they want to look something up later, they look it up by the number, because that is what they wrote down.

The third is speedrunning. The community around fast Wordle play measures things like first-guess solves, the shortest total time across a set of puzzles, and perfect runs across a whole week. Those records are cited by puzzle number because the puzzle number is what pins a run to a specific board. "Solved 1,684 in one guess" is a claim you can check against a single puzzle. "Solved a puzzle in one guess" is not a claim at all.

The fourth, and by far the most common, is the old screenshot. People scroll back through years of posts and find a grid they shared. The grid has colours on it but no date, and often no date in the caption either. The only number in it is the puzzle number, printed in the corner of the board by the game. Ten seconds with the lookup tool above turns that orphan image into a solved puzzle: you get the word, the date, and the vowel and letter breakdown for that exact board.

There is a fifth reason that is a bit less obvious, and it is about verification. If you keep a table of answers by number, then any claim about a past board becomes checkable rather than arguable. Two people remember the same puzzle differently? Pull up the number. Both memories are wrong, or one is, and the list settles it without either person having to be right. That is a small thing, but it is the kind of small thing that turns a group chat argument into a conversation.

When Wordle Started Repeating Answers

The archive contains 1,994 answers and 1,962 distinct words. Thirty-two words have been used twice. For the first 1,689 puzzles — every single day from 19 June 2021 to 1 February 2026 — the answer had never been used before. Not once. That is a run of four years and seven months without a single duplicate, and it is the reason so many players believe as a rule that Wordle never repeats.

It is no longer true. The break came on puzzle 1,690, on 2 February 2026, and the word was CIGAR: the same word that opened the entire game on 19 June 2021, exactly 1,689 days earlier. There is no cleverer pick. The first recycled answer is the opening answer.

Here are the reused pairs, taken straight from the list, with both dates:

Thirty-two is a small number against 1,994, which is why the change is easy to miss. It is also easy to characterise. Every one of those 32 recycled words was first used between 19 June 2021 and 27 October 2024, so the recycling pulls from the older part of the pool rather than from recent puzzles, and every second appearance happened in 2026, between 2 February and 25 November. Nothing was recycled before 2026, and nothing has been recycled from a word first used after October 2024.

The reason is that the published answer pool was expanded and re-rotated in 2026. The original list was finite and, by early 2026, effectively exhausted. Rather than run dry, the publisher widened the pool and started working through it in a way that brings older words back around, so the game keeps moving and players do not feel they have seen the whole thing through.

One thing the data does not do is explain the editorial process itself. It shows the pattern sharply: a clean break at puzzle 1,690, recycled words drawn from the 2021 to 2024 vintage, and a set of brand-new answers still appearing throughout 2026 — 273 of the 305 answers since puzzle 1,690 are words that had never been used before. The pool was both widened and recycled, not merely looped. How the publisher balances those two things is their editorial business, and this list cannot prove a motive. What it can prove is when the behaviour changed and which words were involved, and that is enough to explain why you might meet a familiar word on a board and doubt the spoiler you are reading.

For a player, the practical change is simple. A word you recognise is no longer evidence that a spoiler is wrong. The check is always the puzzle number. If the number on your screen matches the number in the answer list, the word is right, even if you have seen it before. That is one more reason to track puzzles by number rather than by memory of the words.

The Letter Breakdown the Tool Shows

Every answer card on this page carries a small breakdown, and across the whole list those breakdowns describe the answer pool quite precisely. The list holds 9,970 letters in total. E is the most common of them, appearing 1,080 times, which is 10.83% of every letter in the corpus. J is the rarest at 23 appearances, or 0.23%, followed by Q at 0.28%, X at 0.34% and Z at 0.38%.

Two positions are more informative than the raw counts. X has never once started an answer. It is the only letter that cannot do it, which means an X in the first position of your guess is testing something the rotation has never tested. In the other direction, J, Q and V have never ended an answer. Words ending in those three letters simply do not appear as answers, and if you have ever wanted a quick elimination on a board where the yellows have dried up, that is one to know.

The ends are more predictable than the starts. E ends 395 answers, 19.81% of the list, which makes it the single most common final letter by a wide margin. S starts 300 answers, 15.05% of the list, while Z starts exactly two. So there is one letter that dominates the front of the list and one that dominates the back, and the rarest letter in the whole corpus is also the one you will never see in the first slot.

Two structural facts round it out. 619 answers, 31.04% of the list, contain a letter at least twice. That is not a corner case; it is a third of all puzzles, and it is why duplicate detection belongs in any serious approach to reading a board. A grey tile next to a coloured copy of the same letter does not mean the letter is absent. It means the answer contains exactly one of them, and the answer is usually LEVEL, ERROR, or something close in shape.

And the vowel distribution is strikingly narrow. 1,158 answers, 58.07% of the list, contain exactly two vowels. Only 205 contain three, only 3 contain four, and just 8 contain no vowel at all — words like CRWTH and PHPHT that are almost never guessable in the first place. Two vowels is the default shape of a Wordle answer, which is why a board that has used up its easy vowels often narrows faster than players expect.

The full analysis, with every letter ranked and every distribution broken out, is on the Wordle answer statistics page. The breakdown above is the short version, and it is the version that actually changes how you play.

Puzzle 1 Was CIGAR

The first published Wordle was CIGAR, on 19 June 2021, and it is still the correct answer to puzzle 1 on this list. It is a strange choice to open with, and that was part of the point.

Josh Wardle built the original answer list by hand, working down a list of about thirteen thousand words and striking out anything obscure, rude, or likely to start an argument at a breakfast table. The result was a pool of everyday five-letter words that were recognisable once revealed but not obvious beforehand. CIGAR is squarely in that spirit: an ordinary noun that a lot of players would never put down first, and that everyone recognises the moment it turns green. Opening with it told you something about the game before you had played a single round of it.

The early reputation of the game rested on that careful curation. For the first months, Wordle was a small, self-contained daily habit: one puzzle, one share, one grid. The community that formed around it was small enough that the answers were not going to be spoiled by an algorithm, and the puzzle numbers ran fast enough that you could hold the whole history in your head if you paid attention. The counter started at 1 on the first day and has been climbing ever since, without a single break.

Almost five years later the game handed out CIGAR again, as puzzle 1,690. Which is a neat way for a counter to close its own loop: the number that started it all is also the number that marked the end of the never-repeats rule.

How Puzzle Numbers Fit the Rest of This Site

A number lookup and a date lookup are two views of the same list, and the site keeps them on separate pages on purpose. If you know the number, use the tool above. If you know the date, use the Wordle answer archive, which is organised by date, searchable by word, and filterable by year and by first letter. Neither page tries to do the other's job, and together they cover the two ways people actually remember a puzzle: the number in the corner of the board, and the day it happened.

For the analysis behind the numbers, the answer statistics page is the companion. It counts every letter, every first and last position, every vowel count and every repeated word across the full list, and it is where the harder questions about the pool get answered properly. This page is the fast lookup; that one is the long read.

Two other pages are worth having open while you use either of them. The hint ladder walks from a general nudge down to a near-complete reveal, which is the right stopping point if you want to finish a puzzle rather than look it up. The Wordle solver filters a large candidate list against the colours on your board, and it is the right tool when you already know the shape of the answer and just need the candidates narrowed. If you want to play rather than look things up, the practice game gives you an unlimited board that costs no streak. And if the New York Times version is blocked where you are, this free Wordle alternative plays the same game without the network dependency.

Building a Streak Tracker from This List

If you keep streaks, or want to start keeping them properly, this list is a good foundation for it. The important part is not the lookup tool but the shape of the data underneath it, because a tracker needs to look things up by number rather than by date.

Store each answer as a record with four fields: the puzzle number, the ISO date, the answer word, and your own result. The puzzle number is the primary key, because it is the only one of the four that never changes meaning. The date can shift by a day depending on which time zone you replay it in, the word can legitimately appear twice now that recycling has started, and the result is your data, not the game's. The number is the one field that stays put.

The simplest useful tracker is a streak counter over a list of results. You do not need a database. A line per puzzle, with the number, the guess count and a pass or fail flag, is enough to compute a current streak, a longest streak, a guess distribution and a monthly average. The puzzle number makes the line sortable and makes it easy to find the gap: if your streak broke, the missing number is the puzzle you missed, and this page tells you what the word was and whether it was a hard board.

Adding the answer word to each record pays off more than you would expect. Once you have a few hundred rows, you can group your losses by board shape: losses on puzzles with a repeated letter mean your duplicate handling is weak, losses on puzzles with two vowels mean you are over-filtering, and losses on puzzles that start with S mean you are wasting your first guess. The breakdown this page prints for every puzzle gives you those properties without you having to work them out by hand.

Two warnings. First, do not scrape this page in a loop as if it were an API; it is a static page built to be read, and there is a page for the current day's answer that is the right thing to poll if you need today's word. Second, keep your own copy of anything you care about. The list is regenerated on a schedule, so its contents change as the publisher publishes more puzzles. Anything you archive from it should live in your own file.

What This Dataset Does and Does Not Cover

This list is a record of published answers. It is not a record of guesses, it cannot tell you how many tries a puzzle took you, and it does not know whether you solved it. It also is not a promise about any future puzzle.

The coverage runs from puzzle 1 on 19 June 2021 through puzzle 1,994 on 5 December 2026. The official feed currently runs to 5 December 2026, and the file behind this page is rebuilt by a scheduled job, so the tail of the list moves forward as new answers are released. On the day a new puzzle appears there can be a short window before it is added.

Everything else is verified against the source feed rather than typed in, which matters more than it sounds. A page that prints a puzzle number is making a checkable claim. If an entry here ever disagreed with the word on your own screen, that would be a genuine error, and it is one you can spot in seconds by comparing the two.

One more honest caveat. The answer pool itself is edited over time, so this is a record of a moving target rather than a fixed historical object. A word that was later removed from the pool stays in this list, because the puzzle that used it really did use it. If you want a fuller picture of how the pool behaves rather than what it contains on one day, the statistics page counts all of it, and the archive lets you search the whole run by date, by word, or by year.