Pick your files, tick the steps, press Start. Zippy turns JSON, CSV and Parquet into the format you need, compresses it, and sends it to your own cloud — without your files passing through our servers.
macOS, Windows and Linux · Free for 10 GB a month · No account
An example run on generated sample files. Your results depend on your data.
Convert to another format, compress, upload — tick the ones you need, in any combination. Each works on its own or chains with the next.
JSON, CSV, TSV, NDJSON and Parquet in any sensible direction, plus Excel, XML, Avro and Arrow as inputs. See a preview of the output before you run.
CSV tables are compressed column by column; other files with OpenZL's general compressor. Every run is checked: Zippy decompresses the result and compares SHA-256 before it says done.
Send results to your own S3, GCS, Azure, SFTP or Google Drive. Large files upload in parts, and a failed upload can be retried without reprocessing.
Point Zippy at a folder and every new file runs through a saved workflow. Pause, edit or remove a folder any time.
Every run is listed with its size and result. Failed runs show in red with a Retry that reloads the files and the steps you used.
Train a profile on sample files and Zippy shows what it bought you against the default. It is only applied automatically when it beats the default. macOS and Linux for now.
Nothing goes to us. Cloud credentials stay in your system keychain, optionally behind a passphrase. Anonymous usage counts are on by default and one switch turns them off.
zippy compress, convert, pipeline, watch and more — twelve commands on the same engine, for cron jobs and CI, on macOS and Linux. It shares the app's free allowance.
The app follows your system appearance, and so does this page. One accent colour, no gradients, nothing in the way.
For CSV and TSV files Zippy compresses each column with a codec that suits it. These are public datasets, compressed sizes shown as how many times smaller the file got — higher is better.
| Dataset | Size | zstd -3 | xz -6 | Zippy |
|---|---|---|---|---|
| JHU COVID-19 time series | 5.5 MB | 4.06× | 8.53× | 8.94× |
| Our World in Data COVID-19 (first 20 MB) | 20.0 MB | 9.89× | 14.22× | 16.33× |
| World Bank GDP by country and year | 0.56 MB | 3.26× | 4.24× | 4.81× |
| S&P 500 financials | 0.10 MB | 2.95× | 3.48× | 3.53× |
| Airports | 8.8 MB | 2.86× | 3.96× | 3.45× |
| World cities | 1.3 MB | 3.33× | 4.59× | 3.86× |
| NBA Elo ratings | 18.2 MB | 3.86× | 5.49× | 4.30× |
| ISO country codes (mostly text) | 0.13 MB | 2.92× | 3.54× | 3.01× |
Best result per row in orange. Default settings of each tool on the same machine; every Zippy file was decompressed and compared byte for byte. Zippy also compresses each file the ordinary way and keeps the smaller, so the CSV profile never makes a table bigger than ordinary compression would (a file of a few hundred bytes can still grow by its archive header). Applies to CSV and TSV up to 64 MiB; JSON and other formats compress in the same class as zstd and gzip. It takes a few times longer than the ordinary path.
Every step, destination and watch folder is in the free plan. A license removes the monthly limit.
14-day refund — see the Terms. Teams or volume pricing? hello@zippypro.xyz
No. Everything runs on your machine. Files only leave it when you add an Upload step to a cloud destination you set up yourself. Zippy sends anonymous usage counts — never file names or contents — and you can turn that off in Settings → General.
For CSV and TSV tables up to 64 MiB, Zippy reads the table column by column with OpenZL and compresses each column with a codec that suits it. On the public datasets we tried that came out well ahead of zstd -3 and gzip, and on number-heavy tables ahead of xz as well (see the table above). For everything else, such as JSON or Parquet, its default compression lands in the same class as zstd and gzip: similar sizes at similar speed. On top of that Zippy adds the workflow: convert, compress and upload in one run, watch folders, saved workflows, and profiles you can train on your own data.
CSV and TSV files up to 64 MiB are read column by column, and each column is compressed with a codec that suits it. Zippy also compresses the file the ordinary way and keeps whichever result is smaller, so the CSV profile never makes a table bigger than ordinary compression would. (A file of a few hundred bytes can still grow by a few bytes of archive header.) It takes a few times longer than the ordinary path; Fast mode skips it. On macOS and Linux you can also train a CSV profile on your own tables in Settings → Profiles.
Yes. After each compression Zippy decompresses the result and compares SHA-256 checksums before reporting success.
JSON, CSV, TSV, NDJSON and Parquet in any direction that makes sense, plus Excel, XML, Avro and Arrow as inputs — 32 conversions in all. Converting to Parquet is limited to 500 MB per file because the whole dataset has to fit in memory; CSV and NDJSON stream straight to disk, so use those for bigger files.
Up to 10 GB per file, 500 files and 50 GB per run. Large files are split into 64 MiB parts inside an ordinary .zl file, so memory use stays bounded.
You get 10 GB of processing a month with every conversion, upload and watch folder unlocked. The size of each file you process counts once per run, however many steps you chose. Decompressing is always free. The file that crosses the limit still finishes; after that the rest wait until the allowance resets on the 1st.
No monthly limit, using trained profiles when you compress, and the MCP server for AI tools. Everything else is already in the free plan.
Zippy runs OpenZL's trainer on sample files you choose, for up to three minutes, then compresses those samples with and without the new profile and shows you both results. If a profile doesn't beat the default, Zippy tells you and never applies it automatically. Training is free; using a profile when you compress needs a license. Training runs on macOS and Linux; it isn't in the Windows app yet.
AWS S3 (and S3-compatible services such as MinIO, Backblaze B2 and Cloudflare R2), Google Cloud Storage, Azure Blob Storage, SFTP, and Google Drive (one file at a time). Credentials live in your operating system's keychain and can be locked behind a passphrase. They are never sent to us.
They run while Zippy is open. If watch folders are active, closing the window asks before quitting. Turn on Launch at startup in Settings to start Zippy when you log in.
Use Decompress on the Process screen, or run zippy decompress in a terminal. Archives made by Zippy 1.2 or newer open in any later version. A .czl file is an older chunked archive from before 1.2; Zippy still opens those.
macOS, Windows (x64 installer) and Linux (.deb and AppImage).
Converting and compressing work fully offline. Uploads need a connection to your destination, and your license needs to be activated online and re-checked from time to time.
Yes. Deactivate it on the old machine in Settings → General and activate it on the new one. A license covers up to two of your own machines. See the Terms for the 14-day refund policy.