Files, Binary & Compression
Pick File A and File B, then run Compare to see every byte range that differs between them.
Drop a file here, or choose one.
Drop a file here, or choose one.
Files are compared locally in your browser; they are not uploaded.
Change a file and compare again to update the result.
Choose File A and File B, then run Compare.
You reach for a binary file comparator like this one when a checksum only tells you two files differ, not where: a re-encoded asset, a patched firmware image, two builds of the same binary. To compare binary files here, both are read locally and every differing byte range is reported by its offset and length, instead of a single pass or fail verdict.
This is not a text diff. It never tries to detect an inserted or deleted byte and re-align everything after it — it compares offset N of File A against offset N of File B, the same policy `cmp` and most hex-compare tools use, and the one that keeps a comparison predictable on formats that have no concept of a line.
A byte-level answer to what changed, not just whether it changed.
Instead of one match-or-mismatch verdict, the block list gives each differing region an offset and a length, and clicking one jumps the hex view straight to it with a few bytes of context on each side.
When File A and File B are not the same size, the extra trailing bytes of the longer one are reported as their own block instead of being folded into the last changed region.
Drop a file onto either slot or use Choose file. Swap A / B exchanges the two — instantly, without recomputing, if a result is already showing — and Load example fills in a small pair with both a changed region and a length difference.
The comparison does not start the moment both files are picked — press Compare, or Ctrl/Cmd+Enter, so a large pair never runs before you are ready. Changing either file marks the previous result as stale.
Differences only shows just the diff block list; Browse all bytes lets you page through the whole file instead. Either way, Jump to offset takes a decimal or 0x-prefixed hexadecimal value.
Both files are read in matching 1 MiB chunks and compared at the same offsets. A block is a run of strictly consecutive differing bytes, so the result is fully deterministic.
Every differing region appears with its offset, length and kind, and selecting one moves the hex view there and enables exporting that block as a fragment.
Copy or download a plain-text report listing both file sizes, the byte and block counts, and every block's offset and length.
Differences only jumps between diff blocks; Browse all bytes pages through the entire file at a chosen number of bytes per row, for a plain read-through.
The comparison itself always starts at offset 0 of the shorter file, but Jump to offset moves the hex view to any position you give it — decimal or 0x-prefixed hex — so you can inspect a specific region without scrolling.
It compares up to the length of the shorter file, and the extra trailing bytes of the longer one are reported as their own block — labelled as extra bytes in A or in B — rather than being counted as ordinary value differences.
This tool follows the same strict, offset-aligned policy as `cmp`: byte N of one file against byte N of the other, no re-alignment. A text-oriented `diff` instead tries to detect inserted or deleted lines, which does not translate cleanly to arbitrary binary data.
Tools named for reverse engineering, such as BinDiff, compare disassembled code structure to find equivalent functions across builds. This tool works at the raw byte level on any two files and does not disassemble or interpret their contents.
No. Both files are read and compared locally in your browser tab. The report you can copy or download only contains offsets, lengths and file sizes, never the file contents.
Still have a question?
Load the example pair first to see a changed block and a length difference reported side by side, then compare binary files of your own.
Everything runs locally in your browser.