Wordic Games logoWordicGames

Play Semantic Search Online

A close guess can look nothing like the answer. Try SEA and you might move toward OCEAN; try a word that merely shares the same letters and you may get nowhere. Semantic Search ranks a small set of related concepts by meaning. Guess the hidden word.

Loading today's puzzle...

How to play Semantic Search

  1. Read the broad theme and enter a common English word. The answer is one word.
  2. If the word is in this puzzle's curated vocabulary, you get a rank. Number 1 is the answer; smaller numbers mean a closer editorial association.
  3. Use your best-ranked guesses to narrow the topic. There is no guess limit or timer.
  4. Words outside the curated list receive a clear message rather than a fabricated rank. Progress stays on this device for the UTC day.

Meaning beats letter patterns here

If the target is OCEAN, SEA is near the top. WATER and WAVE also point in the right direction. CHAIR might be a common word, but it belongs to a very different subject. Counting shared letters or matching word lengths will not help.

This edition is a deliberately small, hand-ranked semantic vocabulary. It does not run a language model or claim to compute embedding distances. Only words in the published daily concept set can be ranked. If you want a full open-vocabulary embedding game, this is not one yet.

Three ways a rank can steer you

Imagine the theme is “The natural world.” You guess RIVER and receive a middling rank. Try WATER next, then SEA. If the rank improves each time, search for a body of water rather than another land feature.

On the arts board, PIANO may rank fairly close, but SONG is closer to MUSIC. The improvement tells you that the answer is broader than a single instrument. Test another general term before cycling through instrument names.

On the outdoors board, TREE is related to GARDEN, yet YARD and FLOWER are closer in this editorial list. A rank measures the association in this puzzle, not a universal scientific distance between two words. Use several guesses, not just one, to infer the direction.

What the ranking cannot tell you

Different people would rank CORAL and SHARK differently against OCEAN. A hand-authored order is a game design choice, not an objective measurement. We expose that choice rather than labeling it an AI-generated similarity score.

There is no penalty for exploring the vocabulary. The closest-guess display sorts your successful entries so you can see which direction helped. An out-of-list guess does not count against you. Try broad topic words first, then move toward more specific terms if the rank falls.

An opening that covers the space

Start with the most general word the theme allows. On “The natural world,” WATER is better than CORAL as a first guess because it sits near the center of the topic. A broad miss still teaches you; a narrow miss teaches you only that one niche was wrong.

Then bracket the answer. If SEA ranks higher than WATER, the target is probably a body of water, so test OCEAN, LAKE, and RIVER rather than guessing more adjectives. If your second guess ranks worse, pivot to the other side of the theme entirely. Two ranks on opposite sides of the list triangulate faster than five guesses inside the same niche.

Save near-synonyms for last. When the top two candidate words seem interchangeable, submit both before inventing a third candidate. The rank difference between them is itself information.

How this compares

GameThe key difference
Semantic SearchHand-ranked concept associations; no letter-position hints.
LexleLetter placement feedback with a fixed six-guess budget.
Connections GroupsGroup sixteen given words by links rather than guessing a hidden word.

Questions players ask

Does Semantic Search use AI embeddings?

No. This compact edition uses an explicitly hand-ranked vocabulary of thirty related words per answer. Unknown words receive no rank.

Is a rank of 1 the answer?

Yes. The answer always has rank 1. Lower numbers are closer to it in the daily curated concept list.

Why does my valid word have no rank?

The game recognizes only the thirty published concept guesses for each daily puzzle plus the answer. It does not generate a score for arbitrary dictionary words.

Can I try unlimited guesses?

Yes. There is no guess limit. Try broad theme words, compare their ranks, then follow the strongest association.

Do spelling or word length affect the rank?

No. Ranks reflect meaning alone. A five-letter guess gets no advantage over a three-letter guess, and a word sharing letters with the answer is not closer by spelling.

Keep playing

Play today's board above, then try a different type of deduction in Lexle or practice scanning letter grids in Word Search. For the rules of the original format, read Contexto's original game. Unlike open-vocabulary embedding games, this edition has an intentionally limited hand-ranked word set.

You might also enjoy

Word tools

Related articles