The most powerful
compression on Earth
LZMA hasn't changed since 1998. We built something that understands your data — 151 specialized strategies that compress what other tools can't.
Built by a solo founder. Open source. No VC. Working for tips. Read the story →
Part of the Stillwater OS ecosystem
The compression problem
You're paying to store data that could be 20-45% smaller. And the tools you're using haven't fundamentally improved in decades.
Data is growing 25% per year
Global data creation hit 120 zettabytes in 2023. Your storage bill is growing faster than your budget.
LZMA hasn't changed since 1998
The best general-purpose compressor is 28 years old. It treats your CSV the same as your JPEG.
One size fits none
Current compressors see bytes, not structure. A CSV has columns. A log has templates. They ignore all of it.
Energy costs are real
Datacenters use 1-2% of global electricity. Less storage means less power, less cooling, less CO2.
How PZIP works
Current compressors see bytes. PZIP sees structure. A CSV has columns. A log has templates. Your JSON has a schema. We exploit all of it.
Detect
PZIP identifies your file type — CSV, JSON, logs, text, source code, and more.
Select
From 151 specialized compression strategies, it picks the ones designed for your data.
Compress
Extracts structure and compresses what's left. The result is smaller than any general-purpose compressor.
Verify
Round-trip correctness: decompress and compare byte-for-byte. Every single time.
$ pzip data.csv
Detected: CSV (47 columns, 26K rows)
Strategy: Columnar codec (4 specialized strategies selected)
Original: 4,523,891 bytes
PZIP: 2,749,244 bytes (ratio: 0.608)
LZMA-9: 4,512,003 bytes
Saving: 39.2% vs LZMA-9
RTC: VERIFIED (byte-exact)
Output: data.csv.pz
$ pzip -d data.csv.pz
Decompressed: data.csv (4,523,891 bytes)
RTC: VERIFIEDThe datacenter math
S3 charges $0.023/GB/month. At petabyte scale, every percentage point matters. PZIP delivers 20-45% on the file types that dominate your storage.
Storage Costs
Annual savings on a 10 PB datacenter (15% median improvement on CSV/JSON)
Transfer Speed
Smaller files = faster network transfers, lower bandwidth bills, snappier APIs
Energy & CO2
Less storage = fewer drives = less power = less cooling. Compression is climate action.
Zero Hardware
A drop-in upgrade. No new hardware, no migration. Just better compression.
1 PB stored on S3 = $23,000/month
20% better compression = $4,600/month saved
45% on JSONL-heavy loads = $10,350/month saved
Annual savings (15% median) = $41,400 - $124,200
That's just storage. Add bandwidth, transfer,
and the drives you don't have to buy."How does it work?"
We'd love to tell you everything, but our lawyers said to get the patent first. Think Pied Piper, but we actually ship. And our Weissman Score is real.
"Is this like middle-out compression?"
Better. We go outside-in, inside-out, upside-down, and occasionally diagonal. 46 theoretical papers and zero HBO royalties.
What we CAN tell you:
- 151 specialized strategies — different weapons for different file types
- 46 theoretical papers — built on information physics, not heuristics
- Never-worse guarantee — if we can't beat LZMA, we output LZMA. You literally cannot lose.
Ready to compress smarter?
Upload a file and see the difference. Free during beta — no sign-up, no credit card, no data stored. Just better compression.
3117 W / 0 T / 67 L • 3184 files tested • RTC verified • Patent pending
Help me build PZIP Desktop
PZIP is built by one person. Right now it runs in the cloud. With enough support, I'll build a native desktop app and open source everything.
Compress locally. No upload. No cloud. No limits. Your files never leave your machine.
Web demo stays free. Always.
Native. Offline. Unlimited files.
Full engine on GitHub. MIT license.
Every tip goes directly to building PZIP Desktop.
1980s shareware vibes — free software, honest work, tips welcome