cmix is a lossless data compression program aimed at optimizing compression ratio at the cost of high CPU/memory usage. It gets state of the art results on several compression benchmarks. cmix is free software distributed under the GNU General Public License.

cmix works in Linux, Windows, and Mac OS X. At least 32GB of RAM is recommended to run cmix. Feel free to contact me at if you have any questions.

GitHub repository:


Source Code Release Date Windows Executable November 4, 2023 August 29, 2021 August 1, 2019 March 24, 2019 October 3, 2018 May 5, 2018 October 20, 2017 April 24, 2017 November 7, 2016 July 3, 2016 May 30, 2016 April 8, 2016 November 10, 2015 February 4, 2015 September 2, 2014 August 13, 2014 July 23, 2014 June 27, 2014 May 29, 2014 April 13, 2014


Corpus Original size
Compressed size
Compression time
Memory usage
calgary.tar 3152896 536882 2645.67 21228024
silesia 211938580 28302847
enwik6 1000000 175005 773.73 20939644
enwik8 100000000 14760552 65302.38 22852788
enwik9 1000000000 109877715 621780.5 31650128

Compression and decompression time are symmetric. The compressed size can vary slightly depending on the compiler settings used to build the executable.

External Benchmarks

Silesia Open Source Compression Benchmark

File Original size
Compressed size
dickens 10192446 1810680
mozilla 51220480 6639319
mr 9970564 1827572
nci 33553445 779685
ooffice 6152192 1224794
osdb 10085684 1965139
reymont 6627202 705620
samba 21606400 1596282
sao 7251944 3727109
webster 41458703 4287952
xml 5345280 234827
x-ray 8474240 3503868

Calgary Corpus

File Original size
Compressed size
BIB 111261 17264
BOOK1 768771 174520
BOOK2 610856 106519
GEO 102400 42806
NEWS 377109 76889
OBJ1 21504 7052
OBJ2 246814 40256
PAPER1 53161 10877
PAPER2 82199 17246
PIC 513216 21889
PROGC 39611 8238
PROGL 71646 8851
PROGP 49379 6167
TRANS 93695 10040

Canterbury Corpus

File Original size
Compressed size
alice29.txt 152089 31257
asyoulik.txt 125179 29585
cp.html 24603 4765
fields.c 11150 1921
grammar.lsp 3721 778
kennedy.xls 1029744 7960
lcet10.txt 426754 73742
plrabn12.txt 481861 112726
ptt5 513216 21889
sum 38240 6867
xargs.1 4227 1128


I started working on cmix in December 2013. Most of the ideas I implemented came from the book Data Compression Explained by Matt Mahoney.

cmix uses three main components:

  1. Preprocessing
  2. Model prediction
  3. Context mixing

The preprocessing stage transforms the input data into a form which is more easily compressible. This data is then compressed using a single pass, one bit at a time. cmix generates a probabilistic prediction for each bit and the probability is encoded using arithmetic coding.

cmix uses an ensemble of independent models to predict the probability of each bit in the input stream. The model predictions are combined into a single probability using a context mixing algorithm. The output of the context mixer is refined using an algorithm called secondary symbol estimation (SSE).




cmix uses a transformation on three types of data:

  1. Binary executables
  2. Natural language text
  3. Images

The preprocessor uses separate components for detecting the type of data and actually doing the transformation.

For images and binary executables, I used code for detection and transformation from the open source paq8pxd program.

I wrote my own code for detecting and transforming natural language text. It uses an English dictionary and a word replacing transform. The dictionary comes from the phda Hutter prize entry and is 415,377 bytes.

As seen on the Silesia benchmark, additional preprocessing using the precomp program can improve cmix compression on some files.

Model Prediction

cmix v20 uses a total of 2,090 independent models. There are a variety of different types of models, some specialized for certain types of data such as text, executables, or images. For each bit of input data, each model outputs a single floating point number, representing the probability that the next bit of data will be a 1. The majority of the models come from other open source compression programs: paq8l, paq8pxd, and fxcm.

LSTM Mixer


The byte-level mixer uses long short-term memory (LSTM) trained using backpropagation through time. It uses Adam optimization with layer normalization and learning rate decay. The LSTM forget and input gates are coupled. I created two other projects which compress data using only LSTM: lstm-compress and tensorflow-compress. Their results are posted on the Large Text Compression Benchmark.

Context Mixing


cmix uses a similar neural network architecture to paq8. This architecture is also known as a gated linear network. cmix uses three layers of weights.

Comparison to PAQ8

cmix shares many similarities to the PAQ8 family of compression programs. There are many different branches of PAQ8. Here are some of the major differences between cmix and other PAQ8 variants:


Thanks to AI Grant for funding cmix.

cmix uses ideas and source code from many people in the data compression community. Here are some of the major contributors: