Measured - byte-exact - 2026

Pedulli vs lz4

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

TL;DR - honest
lz4 was designed for extreme decompression throughput and accepts a weaker ratio in exchange. Pedulli is a best-of-N racer - it races xz, zstd, brotli and your data's SRD math, keeps the smallest verified output, and never ships a larger one (worst case +1 byte over its own input). On ratio it is never larger than the best standard codec: it wins outright on structured data and ties the best codec on already-optimal/random data. lz4 wins raw throughput. They serve different jobs.

The measured table

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

Input	lz4 -9	Pedulli (best-of-N)	Δ
1 MiB of zeros	4,184 B	13 B	much smaller (redundant input)
JSON 31 KB	6,238 B	1,265 B	-80%
HTML 161 KB	63,488 B	38,757 B	-39%
Apache logs 3.5 MB	489 KB	217 KB	-55.6%
MP4 master 10 MB	10,156,032 B	9,430,108 B	-711 KB
Random bytes 1 MB	1,048,605 B (+29 B)	1,048,577 B (+1 B)	28 B less overhead

What lz4 does better (honest)

Decompression throughput. Best in class for "decompress in the read path" workloads.
Compression throughput. Hundreds of MB/s. Pedulli's coder race is slower at the algorithm level.
Embedded everywhere. Linux kernel, ZFS, Hadoop, ClickHouse, Kafka all default to lz4 for the fast path.

What Pedulli does that lz4 does not

+1 byte never-worse floor vs lz4's +29 on random data.
Best-of-N racing. Races xz, zstd, brotli and SRD math transforms and keeps the smallest verified output - so the result is never larger than the best standard codec (worst case +1 byte). Wins outright on structured data; ties the best codec on already-optimal/random data.
Web-native + API. In-browser for small inputs, REST API for pipelines.
Zero data retention. GDPR by design, EU-sovereign.

When to use which

Use lz4 when latency dominates and ratio is secondary (in-memory caches, compressed swap, read-path decompression). Use Pedulli when ratio matters: cold storage, log archives, video masters, JSON archival - plus the per-file +1 byte never-worse floor.

Run the free trial Pricing / API All benchmarks

More comparisons: zstd - Snappy - LZO - lzop