Compare Two Texts and See the Differences
Compare two blocks of text and see additions and deletions highlighted line by line, instantly.
Both texts stay on your device
Nothing to delete afterwards
Built for confidential pairs
Works in every modern browser
How it works
- 1
Paste the original
The earlier version, the reference, or the text you were given.
- 2
Paste the changed version
The revision, the returned draft, or the file you suspect differs.
- 3
Read the differences
Switch granularity until the comparison shows what you actually need to see.
Why use this tool
Three levels of granularity
Line, word, and character comparison, because prose and code need different answers from the same input.
Invisible differences surfaced
Trailing spaces, tabs against spaces, and CRLF against LF are shown rather than silently equal or unequal.
Smart quotes caught
A curly quote pasted from a word processor is not the same character as a straight one, and the diff says so.
Whitespace toggle
Ignore formatting-only changes when you want the substance, or include them when the formatting is the point.
Nothing is uploaded
Both texts are compared in your browser — two contract versions never reach a server.
Free, no account
No signup, no watermark, no cap on how much you compare.
A diff is an interpretation, not a fact
This is the thing worth understanding first, because it explains most of the moments a comparison seems wrong. Given two versions of a text, there is no single correct list of changes between them. Many different edit scripts — sequences of deletions and insertions — transform the first into the second, and they are all equally valid as descriptions of the result. A diff algorithm picks one, and it picks by minimising the number of operations, which is a reasonable heuristic and not the same thing as reconstructing what the author actually did. So the output is a hypothesis about what changed, chosen because it is short, and it can be short and still misleading. Reading it as a report of the edit rather than as the edit itself is what stops you arguing with the tool.
Moving a paragraph looks exactly like deleting and rewriting it
This is the clearest case where the minimal answer misses the intent. Take a document, move one paragraph from the middle to the end, change nothing else. A standard diff reports that paragraph as removed at its old position and added at its new one — twice the visual noise of the actual edit, and no indication anywhere that it is the same text. To the algorithm nothing was moved, because movement is not an operation it has; there is only delete and insert. The practical consequence is that reorganising a document produces a diff that looks enormous while the wording is untouched, and a genuine rewrite buried in a reorganisation is nearly impossible to spot. When you know a section moved, compare the two sections directly instead of trusting the whole-document view.
The granularity you choose changes the answer
A line diff, a word diff, and a character diff of the same two texts tell you different things, and picking the wrong one is why comparisons are sometimes useless. Line comparison is right for code and for anything where a line is a meaningful unit; it is close to useless for prose, because editing one word in a wrapped paragraph rewrites every line after it, and the diff reports the whole paragraph as changed. Word comparison is the correct default for writing — it finds the one adjective that moved. Character comparison is finer still and mostly useful for short strings: an identifier, a URL, a key where one transposed digit matters. If the result looks like everything changed, the granularity is usually wrong before the text is.
Two texts that look identical and are not
The complaint usually runs the other way, but this case is more insidious: the diff insists something differs and you cannot see it. There is almost always a character involved that has no visible shape. A trailing space at the end of a line is the most common. Tabs against spaces look the same at the same width. A file edited on Windows carries CRLF line endings where the other has LF, which makes every single line differ. Text pasted from a word processor brings curly quotation marks and en dashes that are different characters from the straight quote and hyphen you typed. And a non-breaking space, or a zero-width character carried in from a web page, is invisible by definition. A comparison that marks whitespace and reports line endings turns an inexplicable difference into an obvious one.
A comparison cannot tell you which version is right
It is worth being clear about what a two-way diff does not provide. It shows that two texts differ and where, and it says nothing about which is newer, which is authoritative, or which change was deliberate. That question needs a third input — the version both sides started from — which is what makes a three-way merge possible and a two-way comparison merely informative. This matters when two people edited the same document separately and you are trying to combine their work. A diff between their two copies shows you every place they differ, including places where one of them made a change and the other simply did not, which look identical in the output. Deciding requires knowing the original, so find it before you start reconciling.
Why comparing locally matters for this kind of text
Consider the pairs of documents people actually compare. Two versions of a contract, before and after the other side's revisions. A draft returned by a client with changes they did not mark. Two configuration files, one of which is production. An employment agreement against a template. A policy document against last year's. Comparison is inherently something you do to important documents, usually at a moment when the difference has consequences — and it requires pasting both complete versions, not an excerpt. A hosted comparison tool therefore receives twice the sensitive material of a single-text tool. This one compares in the page and transmits nothing, which you can confirm in the Network tab of your developer tools.
Common mistakes to avoid
- Reading a diff as a record of what the author did. The algorithm picks the shortest edit script, not the true one, so a short explanation of the change is not necessarily the right one.
- Trusting a whole-document diff after a reorganisation. A moved paragraph shows as deleted in one place and added in another, so the output looks enormous while the wording is untouched.
- Using line comparison on prose. Changing one word in a wrapped paragraph rewrites every line after it, so a line diff reports the whole paragraph as changed — use word level for writing.
- Assuming an unexplainable difference is a bug in the tool. Look for a trailing space, tabs against spaces, CRLF against LF, or curly quotes pasted from a word processor.
- Pasting two contract versions into a hosted comparison tool. Comparison requires both complete documents, so you are sending twice the sensitive material; compare locally.
How it compares
| Aspect | This tool | Online diff sites | A desktop diff tool |
|---|---|---|---|
| Content sent to a server | Never | Usually yes | No |
| Line, word, and character modes | All three | Usually line only | Usually all three |
| Whitespace made visible | Yes | Rarely | Yes |
| Setup required | None | None | Install and configure |
| Account or signup | Not needed | Often required | Not needed |
| Price | Free | Free / paid tiers | Free / paid |
Features
Side-by-side and inline views
Two columns for reviewing, one merged column for reading through a change.
Line, word, and character modes
The same comparison at three resolutions, switchable without re-pasting.
Whitespace made visible
Trailing spaces and tab characters are marked, so a difference you cannot see is still one you can find.
Unicode-aware comparison
Arabic, Turkish, and accented text compare correctly rather than by byte.
Handles long documents
Whole files compare without an upload or a size cap.
Change summary
How many lines were added, removed, and modified, before you read a single one.
Nothing to install
No diff tool, no runtime, no dependencies — it runs on the web page.
Arabic and RTL ready
Full interface in eight languages, including right-to-left Arabic.
Secure by default
Served over HTTPS, with no content tracking and no third-party upload.
Who uses it
Lawyers and contract reviewers
Finding the clause that changed in a returned draft that arrived without tracked changes.
Developers
Comparing two configuration files or two responses without committing anything to a repository.
Editors
Seeing exactly which words an author revised between two versions of a piece.
Translators
Checking what changed in an updated source text before revising the translation.
Frequently Asked Questions
No. Everything runs locally in your browser — your text is never uploaded, stored, or shared.
Yes — completely free, with no account and no limits.