Spelling Bee Solver

Try the spelling bee solver

Ranks, points and what the solver totals

Bee scoring is simple and the ranks are percentages of the day's maximum, which is why a hive with many long words has a much higher Genius threshold than a sparse one. The solver's point total is for its own word list, which is much longer than the Times', so use the answers page for the real maximum. The example column uses 2 September 2026, a 138-point day on the Times' list.

Why the solver's list and the official list differ

The solver searches the dwyl english-words list, split by centre letter so the page downloads only the slice it needs. That list was built to be broad, and it shows. We ran the solver on the 40 hives from 1 September to 10 October 2026 and compared it with the Times' own answer list for each day. On average it returned 188 words where the game accepted 47 — about 143 extra words a day — and the extras include archaic words, technical terms and proper nouns written in lower case: CAIRO, ANNIE, ALLAH and FILIPINA all appear.

It also misses words. On 33 of the 40 days at least one accepted word was absent from the solver's list — 64 words in all, mostly recent or informal ones such as NAAN, ACAI, ROMCOM, PFFT and TELEHEALTH. So the solver is a generous search, not a mirror of the game. For the official list of any day HintSheet has published, use the answers page or the archive, which show the Times' own words.

One hive makes the difference concrete. On 10 October 2026 (centre C; outer A, I, M, O, R, T) the Times accepted 63 words worth 297 points, with one pangram, AROMATIC. The solver, given the same seven letters, returns 333 words worth 1,961 points and nine "pangrams" — AROMATIC plus ACROAMATIC, AMORITIC, COMICOCRATIC, MACROTIA, MICROCOAT, MICROTIA, TIMOCRATIC and TRIATOMIC — and it lacks four words the game accepted that day: ACAI, CAMI, CAMO and ROMCOM. Its point total would put Genius at about 1,373 points, more than four times the real threshold of 208. That is why the solver's total is a curiosity and the answers page's total is the one to play against.

Getting the most from the tool

A few habits make the solver useful rather than overwhelming:

• Enter the centre letter in its own box — it is outlined in gold — then the six outer letters in any order. Order does not affect the result, and a repeated or extra letter is ignored.
• Look at the pangram count before the words. The solver's pangrams can include words the game won't take, and in our comparison they contained every official pangram but one (KNEECAPPED, 30 September); knowing roughly how long the pangram is helps you find it by hand.
• Read by length, starting short. The four-letter section is where the familiar words are; the long sections are where the dictionary's oddities pile up.
• Treat any word you don't recognise as probably rejected. The game's list leaves out obscure words and proper nouns, and the solver's does not.
• Check a past hive against its answer page in the archive to see which kinds of word the Times actually accepted that day.

How the search works

The rules are mechanical and the search is exact for its dictionary. A word is kept when it is at least four letters long, uses only the seven hive letters — any of them any number of times — and contains the centre letter at least once. The solver loads the word file for your centre letter once, keeps the words that pass, groups them by length and marks any word that uses all seven letters as a pangram. Points follow the game's rules: one for a four-letter word, one per letter for longer words, and seven extra for a pangram. Nothing you type is sent anywhere; the check runs in the page.

Because letters may repeat, the possible letter strings are endless, but the dictionary bounds the search: only words in the file are tested. For hives you invent yourself the tool is a quick test of whether they are any good, with one caveat: its counts run roughly four times the Times' list, so a hive that looks rich here may be thin in the game.

Finding the pangram by hand before you look

The pangram is worth more than any other word and it is also the most findable, because it has to use all seven letters and that is a strong constraint. Before you type the hive into the solver, try this for three minutes. Write the seven letters with the centre letter in the middle, then look for the letter combinations that English words are built from: common endings first — ING, ION, ED, ER, ITY — and see which of them the hive can spell. In the Times' hives from 1 September to 10 October 2026, seven contained I, N and G, and five of those had an -ING pangram (CRIMPING, VETOING, WHOOPING among them). Prefixes work the same way: UN-, RE- and DE- each fix two letters at the front.

The other reliable route is the letter that is hardest to use. If the hive has a Q, a Z, a J or an X, the pangram must contain it, and the number of words that contain a Z and six specific other letters is small enough to think through: ANALYZED and DIZZYINGLY were the Z pangrams in our window, JELLYBEAN the J one, and ACQUAINT and EQUINOX the Q ones.

When both routes fail, the solver's pangram list is the next step. On some days it matches the game exactly — on 2 September 2026 it offered FLIPPANT and PLAINTIFF, the Times' two pangrams — but on 30 September it lacked KNEECAPPED, so a missing pangram there does not mean there isn't one.

How the official list drifts, and what the solver does about it

The Times' list changes as the language does — recent words like ROMCOM, ENBY and TELEHEALTH were accepted in our window — and the game invites corrections at buzzwords@nytimes.com. The solver's dictionary does not follow it. It is a fixed public-domain list, so it lags on new informal words and keeps thousands of archaic and technical ones the Times will never accept.

In practice that means two kinds of error, both measurable. Extra words: about 143 a day in our 40-hive comparison, most of them obvious on sight. Missing words: 64 accepted words across 33 of the 40 days, most of them the newest or most informal words on the list. If you find a word the game accepted that the solver lacks, the contact page reaches the editor; please include the date of the hive.

Why word lists disagree, with numbers

The disagreement between any two Bee word lists is not random; it comes from decisions every list-maker has to take. The first is breadth. A list built for word games or for completeness includes rare words — the dwyl list has ACTA, AMIC and hundreds like them for a single hive — while the Times leaves out words it judges obscure. The second is proper nouns: the Times rejects them outright (CAIRO, ANNIE), but a list assembled from many sources can carry them in lower case. The third is recency: the Times adds informal words such as NOOB and ENBY that older lists lack. And the fourth is spelling: the Times accepted ANEMIC and rejected ANAEMIC on the same day.

Put numerically, for the 40 hives from 1 September to 10 October 2026, the Times accepted between 18 and 79 words a day (46.5 on average) and the solver found between 42 and 377 (188 on average). If your only aim is Genius, the four- and five-letter sections of the solver's list are the ones to read, since they hold more of the familiar words and fewer of the oddities — and the answers page will tell you which ones actually count. For any day HintSheet has published since 1 September 2026, the archive carries the Times' exact list, so you can always check the solver against the real thing.

Frequently asked questions