Paste two versions of anything — code, config, prose, logs — and see exactly what changed. Lines shown in red are removed, lines in green are added, and within changed lines, the specific words that differ are highlighted. Runs entirely in your browser — no files uploaded, no history kept.
Longest Common Subsequence algorithm. It finds the biggest set of lines shared between both inputs in order, then everything else is an insertion (green) or deletion (red). Within changed lines, a second pass at word level picks out which parts actually differ.
Split view is easier when there are few changes scattered through long files. Unified view is easier when changes cluster and you want to read them top-to-bottom. Both show the same information — different layout.
Treats lines with only whitespace differences as equal. Useful when one side was reformatted but the logical content is the same. Won't hide whitespace-only lines — they still appear, just not as changes.
For performance, we cap each side at 2,000 lines. The algorithm is O(N²) in memory, so larger inputs would freeze your tab. If your file is bigger, split it into chunks or use git diff locally.
Yes — the "Copy patch" button produces a unified diff (.diff or .patch format) that's compatible with patch, git apply, and most diff viewers. You can also download it as a file.
This tool is text-only. Binary diffing is a different kind of problem and usually needs specialized tools. But this works great on source code, prose, configs, logs, JSON — anything you could reasonably paste into a textarea.