Group anagrams online

It keeps the original spelling of each word and the order in which words first appeared, so the result remains easy to compare with the source list. Use it to organize word-game candidates, check vocabulary collections, prepare exercises, or add a predictable grouping step to a text-processing workflow without writing and maintaining your own matching logic.

How anagram groups are identified

Two words belong to the same anagram group when their letters can be rearranged to produce the same case-insensitive character sequence. The tool creates a stable signature for every submitted word by converting it to lowercase, separating its characters, sorting those characters, and joining them again. Words with an identical signature enter the same cluster. For example, “Listen”, “silent”, and “enlist” share one signature even though their capitalization and letter order differ. The returned values keep the spelling supplied in the request; normalization is used only for comparison. Repeated letters matter, so “tap” does not match “tape,” and “pool” does not match “pol.” Punctuation, spaces, and accented characters are treated as characters rather than silently removed. This literal rule makes the result deterministic and prevents hidden cleanup from changing the meaning of the input. If a collection needs trimming, accent folding, or punctuation removal, perform that normalization before sending the words.

Preparing input and reading the result

Send the words field as an array containing at least one non-empty string. Each distinct letter signature produces one array inside the returned groups array. Groups appear in the order in which their first member occurs, and words inside each group retain their original input order. This makes results stable between runs and useful in tests, exports, and review screens. A word without an anagram companion still appears as a one-item group, because omitting it would lose information from the source list. The comparison is case-insensitive, but the output does not rewrite capitalization: “Tea” and “eat” are grouped together and returned exactly that way. The capability rejects a missing words field, a non-array value, an empty list, empty strings, non-string members, and requests beyond the published item limit. These errors are input errors rather than partial results, so a caller can correct the complete collection instead of receiving a group set built from silently discarded values.

Using deterministic grouping in a workflow

Anagram clustering is useful anywhere a collection must be partitioned by letter composition. A word-game service can group a dictionary subset before presenting possible plays. A teacher can build exercise sets that ask learners to recognize rearrangements. A data-quality pipeline can expose entries that differ only by character order, while keeping unmatched entries visible for inspection. Because the algorithm uses no network access, model inference, randomness, or current time, the same input always produces the same JSON result. That property makes the capability suitable for automated jobs whose outputs are compared, cached, or audited later. Processing runs locally in the browser on the free page, while API requests use the same pure solver for $0.002 per request. For very large dictionaries, split work into bounded batches and decide whether groups spanning batch boundaries must be merged by your application. For ordinary lists, one request gives a complete set of clusters ready to display or process further.