Measured - byte-exact - 2026

Pedulli vs zpaq

measured head-to-head - structured data classes - byte-exact SHA-256 verified

TL;DR - honest
zpaq is a context-mixing archiver: top-tier ratios on natural-language text, but very slow. Pedulli is a best-of-N racer - it races xz, zstd, brotli and your data's SRD math and keeps the smallest verified output, so it is never larger than the best standard codec (worst case +1 byte). It wins outright on JSON, periodic, sparse and MP4, ties the best codec on already-optimal data, and adds +1 byte where zpaq adds 88. On long English prose, zpaq's context-mixing reaches a tighter ratio than the standard codecs the Orchestrator races - pair the two when that single class is the priority.

The measured table

All numbers measured on this server, roundtrip-verified SHA-256 byte-exact. Proofs available on request.

Input	zpaq -m5	Pedulli (best-of-N)	Δ
1 MiB of zeros	187 B	13 B	much smaller (redundant input)
JSON 31 KB	2,341 B	1,265 B	-46%
Natural-language prose 500 KB	~130 KB	~130 KB	tie - selects best codec
HTML 161 KB	33,492 B	33,492 B	tie - selects best codec
Apache logs 3.5 MB	198 KB	198 KB	tie - selects best codec
MP4 master 10 MB	9,455,920 B	9,430,108 B	-25 KB
Random bytes 1 MB	1,048,664 B (+88 B)	1,048,577 B (+1 B)	87 B less overhead

Where zpaq is strong (honest)

Context-mixing on natural-language text. Among the strongest published ratios on long English prose - a tighter ratio than the standard codecs the Orchestrator races on that one class. Pair zpaq with the Orchestrator when long prose is the priority.
Built-in deduplication archiver. zpaq is both a codec and an incremental archive format.

What Pedulli does that zpaq does not

+1 byte never-worse floor vs zpaq's +88 on random data.
Best-of-N racing. Races xz, zstd, brotli and your data's SRD math, keeps the smallest verified output - never larger than the best standard codec (worst case +1 byte). Wins outright on JSON, periodic, sparse and MP4.
Much faster on the same data. zpaq -m5 takes minutes per MB; Pedulli's race is in a practical service tier.
Web-native + API + chunked streaming. zpaq is CLI only.
Zero data retention. GDPR by design, EU-sovereign.

When to use which

Use zpaq for one-off archival of large English-prose corpora where ratio matters more than time. Use Pedulli for JSON, MP4, periodic, logs and streaming pipelines - plus the per-file +1 byte never-worse floor.

Run the free trial Pricing / API All benchmarks

More comparisons: xz - 7-Zip - bzip2 - lrzip