Right letter, right place
Enter the number of letters that match the answer in the same position.
Word500 helper
Use the three totals to choose your next guess
Enter each five-letter guess with its green and yellow counts. The solver keeps every word that could produce those totals, then ranks guesses that separate the shortlist.
Play today’s puzzle on the official Word500 siteYour completed guesses
Recreate the feedback
What still fits
Solver results
Ranked next guesses
Lower average and worst-case shortlists are better. Entropy measures how evenly a guess separates the remaining words.
Candidate words
The list shown here is shortened, but the ranking used the full candidate set.
No candidates match every row
Check the guess spelling and feedback totals. A duplicate letter can change how yellow matches are counted.
How to use it
Enter totals, not tile positions
After a guess, copy the number of green and yellow letters into one row. You do not need to say which positions matched: Word500 gives aggregate feedback, so the solver checks every possible arrangement that produces the same totals.
Right letter, different place
Enter the remaining matching letters after exact-position matches are removed.
No remaining match
The solver derives this as five minus green minus yellow.
Difficulty rules used by the solver
- Standard: five different letters, without J, Q, X, or Z.
- Standard+: five different letters, with the full alphabet available.
- Advanced: repeated letters and the full alphabet are allowed.
Choose the level of the puzzle you are solving before entering its rows.
How repeated letters are handled
Feedback is calculated duplicate-aware. Exact-position matches are counted first; other letters can count yellow only while an unmatched copy remains in the candidate. This prevents one answer letter from satisfying two copies in a guess.
Work through the three modes
These are original examples, not daily answers. All three modes report totals, so a green count tells you how many positions match, not which positions they are. Record the complete guess and both counts; the solver compares that evidence with each candidate. Only enter a known position when another clue actually supplies it.
Standard: separate candidates that give the same totals
Standard answers use five distinct letters and exclude J, Q, X and Z. A CRANE guess can leave both CAPER and CATER possible: each gives one green, three yellow and one red. The shared totals do not tell you whether P or T is present.
See a move that distinguishes this two-word example
POUND gives zero green, one yellow and four red against CAPER, but zero green, zero yellow and five red against CATER. On this two-answer shortlist it distinguishes them. It is a probe, not a guarantee of winning in one guess, and it is not claimed optimal over a larger dictionary.
Standard+: keep the full alphabet in consideration
Standard+ still uses five distinct answer letters but permits J, Q, X and Z. For example, QUILT fits Standard+ and is excluded from the Standard answer pool. CRANE produces zero green, zero yellow and five red against QUILT. Choosing the wrong difficulty can discard a valid answer before the feedback is even checked.
Try the Standard+ feedback
Compare the remaining candidates after changing the difficulty. Zero overlap with CRANE rules out C, R, A, N and E; it does not identify the order of the other letters.
Advanced: allocate repeated letters once
Advanced allows repeated answer letters. In the original ALLEE versus APPLE example, A in position 1 and E in position 5 are green. Only one L can match APPLE's single L, producing one yellow. The other L and the extra E have no remaining matches, so the totals are two green, one yellow and two red.
Try the repeated-letter example
Do not count both L guesses as yellow. Matching uses the number of copies actually present after exact matches are removed. APPLE remains a candidate under these totals; a distinct-letter mode would incorrectly exclude it.
Compare suggested moves using the displayed objective: entropy measures how feedback splits the shortlist, expected remaining averages the branch sizes, and worst case considers the largest branch. None is a claim about observed player success. The search ranks a bounded set of probes against this independent candidate pool; the official game's accepted words may differ.