zstd is a fast lossless compression algorithm and data compression tool, with command line syntax similar to gzip and xz. It is based on the LZ77 family, with further FSE & huff0 entropy stages.
Key Features
- Fast compression: > 200 MB/s per core in fast modes
- Strong compression: Excellent compression ratios with configurable levels
- Fast decompression: > 500 MB/s per core across all compression settings
- Dictionary compression: Dramatically improves compression on small files
- Multi-threading support: Parallel compression with
-T# option
Basic Usage
zstd [OPTIONS...] [INPUT... | -] [-o OUTPUT]
Compress a file
# Compress with default level (3)
zstd file.txt
# Output: file.txt.zst
# Compress with specific level (1-19)
zstd -5 file.txt
# Compress with maximum compression
zstd -19 file.txt
Decompress a file
# Decompress
zstd -d file.txt.zst
# Output: file.txt
# Alternative syntax
unzstd file.txt.zst
Write to standard output
# Decompress to stdout
zstd -dc file.txt.zst
# Alternative
zstdcat file.txt.zst
Common Options
| Option | Description |
|---|
-# | Compression level (1-19, default: 3) |
-d, --decompress | Decompress files |
-o OUTPUT | Write output to OUTPUT file |
-k, --keep | Keep input files (default) |
--rm | Remove input files after processing |
-f, --force | Overwrite existing files |
-c, --stdout | Write to stdout |
-v, --verbose | Verbose mode |
-q, --quiet | Suppress warnings |
-h, --help | Display help |
Differences from gzip
zstd has several behavioral differences from gzip:
- Source files are preserved by default (use
--rm to remove)
- Progress notifications are shown by default (use
-q to disable)
- Does not accept input from console (but accepts stdin when piped)
- Does not store filename or attributes, only file contents
File Handling
Reading from stdin
# Compress from stdin
cat file.txt | zstd > file.txt.zst
# Using dash for stdin
zstd - < file.txt > file.txt.zst
Recursive compression
# Compress all files in directory
zstd -r directory/
# With specific compression level
zstd -r -10 directory/
Output directory options
# Store files in flat directory
zstd -r --output-dir-flat /output/dir input/
# Preserve directory structure
zstd -r --output-dir-mirror /output/dir input/
Symlink Shortcuts
zstd can be invoked through symlinks with special behavior:
| Symlink | Behavior |
|---|
zstdmt | Compress using all CPU cores (zstd -T0) |
unzstd | Decompress files |
zstdcat | Decompress to stdout |
zcat | Same as zstdcat |
Environment Variables
Only ZSTD_CLEVEL and ZSTD_NBTHREADS are supported for security reasons.
ZSTD_CLEVEL
Set default compression level (1-19):
export ZSTD_CLEVEL=5
zstd file.txt # Uses level 5
ZSTD_NBTHREADS
Set default number of threads:
export ZSTD_NBTHREADS=4
zstd file.txt # Uses 4 threads
-# and -T#).
Concatenation
Multiple .zst files can be concatenated and decompressed as a single file:
cat file1.zst file2.zst file3.zst > combined.zst
zstd -d combined.zst
Integer Suffixes
Many options accept integer suffixes for convenience:
# These are equivalent
zstd -B 1048576 file.txt
zstd -B 1MiB file.txt
zstd -B 1MB file.txt
KiB, Ki, K, KB = × 1,024MiB, Mi, M, MB = × 1,048,576
Parameter Aggregation
CLI supports parameter aggregation:
# These are equivalent
zstd -b1 -e18 -i1 file.txt
zstd -b1e18i1 file.txt