String Similarity Checker

Compare any two strings: Levenshtein edit distance, normalized similarity, Jaccard word overlap, and the longest common substring, computed live as you type.

String A

String B

Levenshtein distance
(edits: insert / delete / substitute)

100%

Normalized similarity
1 - distance / longer length

100%

Jaccard word overlap
shared words / all unique words

Longest common substring
(contiguous characters)

Longest common substring

Show Levenshtein DP matrix (strings up to 25 characters)

Where this is useful: deduplicating near-identical records (names, addresses, product titles), building fuzzy-match lookups (a similarity threshold of 80-90% catches typos without false positives), and comparing LLM outputs against expected answers in evals, where exact string equality is too strict but you still want a quantified "how close". Levenshtein counts character edits, so it is strict about ordering; Jaccard ignores word order entirely, so use it when phrasing may be shuffled. For long free-form text, embedding similarity usually beats both.