Free Text Similarity Checker — Cosine & Jaccard Comparison
Is that rewrite actually different — or just the same paragraph with new words? Paste two texts and get a similarity score two ways: cosine similarity over word frequencies and Jaccard similarity over word sets, plus a shared-word count and a diff-style highlight showing exactly which words are unique to each. Everything runs in your browser.
Text Similarity Checker
Compare two texts — cosine & Jaccard scores plus a word-by-word diff view.
100% Free Forever
No Sign-up
No Credit Card
Private — files never leave your browser
Unlimited Use
1 Your two texts
Best with a paragraph or more in each box — very short texts give noisy scores.
2 Comparison settings
Short glue words rarely carry meaning in a similarity score.
Ignore case ("The" = "the")
Ignore common stopwords (the, and, of…)
Show diff-style word highlight
3 Compare & review
Working…
⚖️ Similarity report
—
Cosine similarity
—
Jaccard similarity
Shared words
—
Unique words (A / B)
🆚 Word-by-word diff (■ only in A · ■ only in B)
Text A
Text B
How it works
- Paste two texts — the original and the rewrite, draft A and draft B, or any pair you want to compare. Live word counts track both.
- Tune the comparison — try a sample pair, set the minimum word length, and toggle case sensitivity, the stopword filter and the diff view.
- Compare — the tool builds word-frequency vectors (cosine similarity) and word sets (Jaccard similarity), counts shared words, and highlights words unique to each text side by side.
Everything runs in your browser — your text never leaves this page.
Everyday uses
- Writers & editors — check whether a rewrite is different enough from the source.
- Teachers — spot assignments that are suspiciously similar to each other.
- Students — compare your paraphrase against the original before submitting.
- Content teams — catch accidental near-duplicate pages that split your SEO.
- Bloggers — verify guest posts aren’t lightly-rewritten copies of existing articles.
Rule of thumb: cosine above 80% usually means near-duplicate content; 40–60% is a heavy paraphrase; under 20% is effectively a different text.