What is the difference between line and word diff?

Line diff compares whole lines and is best for spotting added, removed, or reordered paragraphs and list items. Word diff breaks each line into words to highlight small inline edits, like a changed number or a reworded phrase.

How does the diff algorithm work?

It computes the longest common subsequence between the two token streams, then marks tokens present only in A as deletions, only in B as additions, and shared tokens as unchanged — the same core technique behind git diff.

Is anything sent to a server?

No. The comparison runs entirely in your browser. Both responses stay on your machine, which matters when you are iterating on prompts that contain proprietary content.

Can I use this for regression testing prompts?

Yes — keep a known-good reference output as A, paste the latest run as B, and the diff shows exactly what changed. It is a fast manual check before and after a model upgrade or prompt edit.

Why is whitespace sometimes flagged as a change?

Trailing spaces and blank lines count as content in a diff. Use the trim option to normalise whitespace per line so only meaningful differences are highlighted.

AI Output Diff — Gera Tools

Diff two AI responses

When you iterate on a prompt — change the system message, bump the temperature, upgrade the model — you need to know what actually changed in the output. Eyeballing two long responses misses subtle regressions. This tool runs a real diff between version A and version B and highlights every addition, deletion, and unchanged span.

How it works

The diff computes the longest common subsequence between the two token streams (lines, or words within lines), exactly like git diff. Tokens that appear only in A are deletions, tokens only in B are additions, and shared tokens are unchanged. The result is rendered with colour coding so the differences jump out.

A: The capital is Paris.
B: The capital city is Paris.
       ->  The capital [+city] is Paris.

Tips and notes

Use line-level diff to catch reordered paragraphs or added list items, and word-level to catch a single changed number inside a sentence. Enable trimming to ignore trailing whitespace so only meaningful edits show. For regression testing, keep a golden output in A and paste each new run into B — the highlighted diff is your pass/fail signal.