By FiveLetterWords · Guide updated . The puzzle date is shown separately.

This page always keeps Wordle #683 for May 3, 2023. Check that number against your game before opening a hint. The answer remains below a separate reveal.

Wordle changes at midnight in your local time zone. For the current local puzzle, use today’s hints. This dated page keeps the same puzzle when the day changes.

Our independent hints use the verified puzzle answer. Play the official game at The New York Times.

Hints for Wordle #683

What kind of word

Wiktionary lists it as a noun.

What it means

“A large number of people or things.”

Source: Wiktionary contributors, available under CC BY-SA 4.0. Selected and lightly formatted from Wiktionary; long entries may be shortened.

Fill the gap

“We were beset by a _____ of street vendors who thought we were tourists and would buy their cheap souvenirs.”

Source: Wiktionary contributors, available under CC BY-SA 4.0. Example sentence selected from Wiktionary, with the answer removed.

The middle letter

The middle letter is R.

How it starts

Begins with H.

Reveal the answer and explanation

The answer to Wordle #683 was

HORDE: A wandering troop or gang; especially, a clan or tribe of a nomadic people (originally Tatars) migrating from place to place for the sake of pasturage, plunder, etc.; a predatory multitude.

Source: Wiktionary contributors, available under CC BY-SA 4.0. Selected and lightly formatted from Wiktionary; long entries may be shortened.

Word difficulty: Easy

Why it may be easier: five different letters, at least two vowels and none of Q, Z, X, J, K, V or W.

The word rating describes familiarity and letter features. The solver score below measures how a fixed strategy handles the puzzle; an unusual word can still be easy for that strategy.

Compare past puzzle difficulty →

How difficult was it for the solver?

Solver pressure: 14/100 (easier for the solver). This is a percentile among past puzzles under the same strategy: 0 is easiest and 100 is hardest for the solver.

40 Possible answers after CRANE
111 Possible answers after SLATE
35 Possible answers after ADIEU
3 Guesses in the solver replay

This compares past puzzles under the same solver settings; it does not measure how players performed. Compare more puzzles →

Wordle #683, measured

This section compares HORDE's solver-difficulty percentile, first-guess results for ten familiar openers, and opt-in player reports. Each number identifies its dataset.

Solver difficulty, as a percentile

14/100

This score is a percentile, not an opinion. Under the site's fixed solver settings, HORDE measured harder for the solver than about 14% of all 1,968 recorded answers to date. Its rating is easier for the solver. CRANE, SLATE and ADIEU leave about 62 possible answers.

Solver pressure ranks the answer within the pool; it does not measure player performance. The opener table compares first-guess results, and the contributed ratings report player results.

CRANE, SLATE and ADIEU benchmark

TARSE left the fewest candidates: its first-guess feedback left 14 of the 2,392 words in the site's working answer pool. At the other end, SALET left 174 candidates, about 12 times as many.

First-guess feedback each opener would have received against HORDE, and how many answer-pool words would still match it
OpenerFirst-guess feedbackAnswers still possible
TARSE TARSE 14
ARISE ARISE 25
ADIEU ADIEU 35
CRANE CRANE 40
RAISE RAISE 40
TRACE TRACE 40
AUDIO AUDIO 43
STARE STARE 49
SLATE SLATE 111
SALET SALET 174

“Answers still possible” counts the words in the site's 2,392-word working answer pool whose feedback would match that opener exactly. This is pattern arithmetic, not a simulation. A smaller number means the opener left fewer possible answers after the first guess. It does not describe later turns.

Contributed player ratings

Not enough contributed games to publish player figures for this puzzle. Per-puzzle numbers appear only after at least 10 opt-in games arrive. See how the Observatory collects and publishes player data →

What did you get?

Tap your result to compare it with contributed games. Your choice stays in this browser and is not sent anywhere.

1 Suppressed
2 Suppressed
3 13
4 12
5 Suppressed
6 Suppressed
X Suppressed

The bars pool all 44 contributed games across every completed puzzle. The Observatory does not publish per-puzzle guess distributions below its sample threshold, so this is the broadest available comparison. “Suppressed” marks a low-count cell.

How the solver finds HORDE

The strategy receives only earlier guesses and their feedback. It does not know the answer before choosing a move. Guessing the answer itself on turn one is a lucky one-turn result, not a recommended general strategy.

Normal mode

Our solver: bounded weighted policy. Current-pool replay over 14,855 accepted words. Model through 2026-11-05.

  1. The fixed opener tests its five letters.

    14,855 before → 65 after. This feedback leaves 65 candidates.

    All 65 surviving candidates

    BERME, BIRLE, BIRZE, BORDE, BOREE, BORNE, BURKE, CERGE, CERNE, CORBE, CURIE, CURVE, DIRGE, DIRKE, DOREE, DORJE, EERIE, EYRIE, FIRIE, FORCE, FORGE, FORME, FORZE, FURZE, GERBE, GERLE, GERNE, GORGE, GURGE, HERYE, HIREE, HORDE, HORME, JIRRE, KERNE, KERVE, KURRE, KYRIE, LURVE, MERDE, MERGE, MERLE, MOREE, MORNE, MURRE, NERVE, NORIE, OORIE, OURIE, OWRIE, PERCE, PERLE, PERNE, PERVE, PIRRE, PORGE, PUREE, PURGE, PURRE, RORIE, VERDE, VERGE, VERRE, VERVE, VIRGE

    Decision measures before feedback

    Expected remaining candidates: 414.45. Information: 5.74 bits. These measures describe the decision before its realised feedback.

  2. Test GOLES against the 65 candidates consistent with the earlier rows.

    65 before → 13 after. This feedback leaves 13 candidates.

    All 13 surviving candidates

    BORDE, BORNE, CORBE, DORJE, FORCE, FORME, FORZE, HORDE, HORME, MORNE, NORIE, OORIE, RORIE

    Decision measures before feedback

    Expected remaining candidates: 2.849. Information: 1.18 bits. These measures describe the decision before its realised feedback.

  3. Test BORDE against the 13 candidates consistent with the earlier rows.

    13 before → 1 after. This feedback leaves 1 candidate.

    All 1 surviving candidates

    HORDE

    Decision measures before feedback

    Expected remaining candidates: 4.434. Information: 2.0 bits. These measures describe the decision before its realised feedback.

  4. Submit the only remaining candidate.

    1 before → 1 after. The answer is submitted and solved.

    All 1 surviving candidates

    HORDE

    Decision measures before feedback

    Expected remaining candidates: 1.0. Information: 0.0 bits. These measures describe the decision before its realised feedback.

Solved in 4 guesses. Uniquely identified on turn 3.

Continue this board in the Solver

Hard Mode

Our solver: bounded weighted policy. Current-pool replay over 14,855 accepted words. Model through 2026-11-05.

  1. The fixed opener tests its five letters.

    14,855 before → 65 after. This feedback leaves 65 candidates.

    All 65 surviving candidates

    BERME, BIRLE, BIRZE, BORDE, BOREE, BORNE, BURKE, CERGE, CERNE, CORBE, CURIE, CURVE, DIRGE, DIRKE, DOREE, DORJE, EERIE, EYRIE, FIRIE, FORCE, FORGE, FORME, FORZE, FURZE, GERBE, GERLE, GERNE, GORGE, GURGE, HERYE, HIREE, HORDE, HORME, JIRRE, KERNE, KERVE, KURRE, KYRIE, LURVE, MERDE, MERGE, MERLE, MOREE, MORNE, MURRE, NERVE, NORIE, OORIE, OURIE, OWRIE, PERCE, PERLE, PERNE, PERVE, PIRRE, PORGE, PUREE, PURGE, PURRE, RORIE, VERDE, VERGE, VERRE, VERVE, VIRGE

    Decision measures before feedback

    Expected remaining candidates: 414.45. Information: 5.74 bits. These measures describe the decision before its realised feedback.

  2. Test MOREE against the 65 candidates consistent with the earlier rows.

    65 before → 13 after. This feedback leaves 13 candidates.

    All 13 surviving candidates

    BORDE, BORNE, CORBE, DORJE, FORCE, FORGE, FORZE, GORGE, HORDE, NORIE, OORIE, PORGE, RORIE

    Decision measures before feedback

    Expected remaining candidates: 5.004. Information: 1.106 bits. These measures describe the decision before its realised feedback.

  3. Test FORGE against the 13 candidates consistent with the earlier rows.

    13 before → 8 after. This feedback leaves 8 candidates.

    All 8 surviving candidates

    BORDE, BORNE, CORBE, DORJE, HORDE, NORIE, OORIE, RORIE

    Decision measures before feedback

    Expected remaining candidates: 1.313. Information: 0.464 bits. These measures describe the decision before its realised feedback.

  4. Test BORDE against the 8 candidates consistent with the earlier rows.

    8 before → 1 after. This feedback leaves 1 candidate.

    All 1 surviving candidates

    HORDE

    Decision measures before feedback

    Expected remaining candidates: 1.403. Information: 2.292 bits. These measures describe the decision before its realised feedback.

  5. Submit the only remaining candidate.

    1 before → 1 after. The answer is submitted and solved.

    All 1 surviving candidates

    HORDE

    Decision measures before feedback

    Expected remaining candidates: 1.0. Information: 0.0 bits. These measures describe the decision before its realised feedback.

Solved in 5 guesses. Uniquely identified on turn 4.

Continue this board in the Solver
Historical proven-optimal benchmark: July 2023

Alex Selby’s strategy minimises the mean over its complete 3,158-answer snapshot. A route for one answer is not proof of the shortest possible route for that answer, and is separate from the current weighted model.

Normal mode

Selby’s proven-optimal strategy. July 2023 snapshot: 3,158 equally likely answers; minimum mean guesses.

  1. The fixed opener tests its five letters.

    3,158 before → 15 after. This feedback leaves 15 candidates.

    All 15 surviving candidates

    BORNE, CURVE, DIRGE, EERIE, FORCE, FORGE, GORGE, HIREE, HORDE, MERGE, NERVE, PUREE, PURGE, VERGE, VERVE

    Decision measures before feedback

    Expected remaining candidates: 100.074. Information: 5.888 bits. These measures describe the decision before its realised feedback.

  2. Test GENIO against the 15 candidates consistent with the earlier rows.

    15 before → 2 after. This feedback leaves 2 candidates.

    All 2 surviving candidates

    FORCE, HORDE

    Decision measures before feedback

    Expected remaining candidates: 1.4. Information: 3.507 bits. These measures describe the decision before its realised feedback.

  3. Test FORCE against the 2 candidates consistent with the earlier rows.

    2 before → 1 after. This feedback leaves 1 candidate.

    All 1 surviving candidates

    HORDE

    Decision measures before feedback

    Expected remaining candidates: 1.0. Information: 1.0 bits. These measures describe the decision before its realised feedback.

  4. Submit the only remaining candidate.

    1 before → 1 after. The answer is submitted and solved.

    All 1 surviving candidates

    HORDE

    Decision measures before feedback

    Expected remaining candidates: 1.0. Information: 0.0 bits. These measures describe the decision before its realised feedback.

Solved in 4 guesses. Uniquely identified on turn 3.

Continue this board in the Solver

Hard Mode

Selby’s proven-optimal strategy. July 2023 snapshot: 3,158 equally likely answers; minimum mean guesses.

  1. The fixed opener tests its five letters.

    3,158 before → 15 after. This feedback leaves 15 candidates.

    All 15 surviving candidates

    BORNE, CURVE, DIRGE, EERIE, FORCE, FORGE, GORGE, HIREE, HORDE, MERGE, NERVE, PUREE, PURGE, VERGE, VERVE

    Decision measures before feedback

    Expected remaining candidates: 100.074. Information: 5.888 bits. These measures describe the decision before its realised feedback.

  2. Test VERGE against the 15 candidates consistent with the earlier rows.

    15 before → 3 after. This feedback leaves 3 candidates.

    All 3 surviving candidates

    BORNE, FORCE, HORDE

    Decision measures before feedback

    Expected remaining candidates: 2.333. Information: 2.923 bits. These measures describe the decision before its realised feedback.

  3. Test HORDE against the 3 candidates consistent with the earlier rows.

    3 before → 1 after. The answer is submitted and solved.

    All 1 surviving candidates

    HORDE

    Decision measures before feedback

    Expected remaining candidates: 1.667. Information: 0.918 bits. These measures describe the decision before its realised feedback.

Solved in 3 guesses. Uniquely identified on turn 3.

Continue this board in the Solver

Source: Alex Selby’s Wordle analysis, MIT. Licence and attribution.

How the fixed opener panel performs on this answer

OpenerFirst feedbackSurvivorsNormalHard Mode
TARSE00202 (0 grey, 1 yellow, 2 green)6545
CRANE01002 (0 grey, 1 yellow, 2 green)17344
SLATE00002 (0 grey, 1 yellow, 2 green)50655
ADIEU01010 (0 grey, 1 yellow, 2 green)22733
STARE00012 (0 grey, 1 yellow, 2 green)18355
AUDIO00101 (0 grey, 1 yellow, 2 green)36144
RAISE10002 (0 grey, 1 yellow, 2 green)15155
ROATE12002 (0 grey, 1 yellow, 2 green)3344
SALET00010 (0 grey, 1 yellow, 2 green)84044
Normal mode
1 guess: 0 · 2 guesses: 0 · 3 guesses: 1 · 4 guesses: 5 · 5 guesses: 3 · 6 guesses: 0 · Failed in six: 0
Hard Mode
1 guess: 0 · 2 guesses: 0 · 3 guesses: 1 · 4 guesses: 4 · 5 guesses: 4 · 6 guesses: 0 · Failed in six: 0

The nine openers use the same pool and continuation policy. Across both modes: 3–5 guesses among successes; 0 of 18 configurations failed. This is performance on this answer, not an overall opener ranking.

121 openers set up a two-turn opportunity; examples include HORSE, HEADS, HIRED, DROVE, SHORE, HOPED, BORED, SHADE, DRONE, HARDY. After their feedback, this answer is among the most likely survivors with at least half the model probability. This is an opportunity, not a guarantee. An opener that immediately solves is excluded.

Why this answer can be difficult

With TARSE, this policy takes 4 guesses in normal mode and 5 in Hard Mode. These are matched simulations, not reported player averages.

Vocabulary evidence and voluntary player results are reported separately from these solver observations.

A lesson from this route

This route identifies the answer on turn 3. Identifying a unique candidate and submitting it are separate actions.

Try another opener or a different move

How this is calculated

The current policy checks all remaining candidates and up to 128 fixed probes. At six or fewer candidates it uses bounded lookahead, preferring modelled six-turn success, then speed among successes. Larger states use expected survivors as a heuristic. These are recommendations, not proofs of optimal play. Hard Mode retains green positions and revealed letter counts; it allows grey letters and a yellow in the same position. Candidate-only play is stricter.

Methods and limitations

Common questions

What was the Wordle answer for May 3, 2023?
The Wordle answer for May 3, 2023 (puzzle #683) was HORDE.
What does HORDE mean?
HORDE: A wandering troop or gang; especially, a clan or tribe of a nomadic people (originally Tatars) migrating from place to place for the sake of pasturage, plunder, etc.; a predatory multitude.
Why was Wordle #683 tricky?
HORDE is rated easy for estimated word difficulty. Why it may be easier: five different letters, at least two vowels and none of Q, Z, X, J, K, V or W.

Reported player results

Voluntary self-reported sample, not representative of all players. Solver scores do not measure human difficulty. Small or complementary cells stay private.

Wordle #683 · May 3, 2023

Not enough publishable evidence yet. The minimum is 10 reports per protected group; complementary suppression can also withhold a larger group.

Methods, privacy rules and public data. You can optionally contribute a finished result below.

Contribute your Wordle result

Paste the share text from Wordle. After you opt in, the server checks its colours against this puzzle. A grid does not tell us your guesses or prove how you played.

Results appear only after the puzzle finishes worldwide and the privacy threshold is met. Small breakdowns stay hidden. How contributions are used.

Keep solving and playing