Choose a version:
21% The original file has 269933 bytes (263.6k) and is available from the project website.
There you can find the official minified version, too, which brings down the size to 55956 bytes (54.6k, 21%).

After GZIP compression these minified files vary in size:
Microsoft
  25923 bytes (25.3k)
CDN, direct link
Boot
  23772 bytes (23.2k)
CDN, direct link
cdnjs
  20895 bytes (20.4k)
CDN, direct link
gzip -6 (default)
  20738 bytes (20.3k)
local copy
cdnhttps
  20716 bytes (20.2k)
CDN, direct link
gzip -9
  20695 bytes (20.2k)
local copy
unpkg
  20692 bytes (20.2k)
CDN, direct link
jsdelivr
  20684 bytes (20.2k)
CDN, direct link
libdeflate -12
  20052 bytes (19.6k)
local copy
7zip -mx=9 -tgzip
  20046 bytes (19.6k)
local copy
kzip -s0 -rn -b2
  20036 bytes (19.6k)
local copy
pigz -11 -n
  20014 bytes (19.5k)
local copy
Zopfli
  19987 bytes (19.5k)
local copy
Zopfli (defluff)
  19986 bytes (19.5k)
local copy

perma-link to the smallest file on my server:
http://minime.stephan-brumme.com/files/knockout/knockout-3.3.0.min.js

You will automatically get the smallest Knockout 3.3.0 file, ETag caching is available and
if your browser doesn't support GZIP decompression then the uncompressed version will be sent.

Currently best Zopfli settings

Save 697 bytes by using my Knockout 3.3.0 Zopfli version instead of the best available CDN (3.49% smaller than jsdelivr, 19987 vs. 20684 bytes):
You can use my super-compressed files for whatever purpose you like as long as you respect the library's original license agreement.
There are no restrictions from my side - but please avoid hot-linking if you run a high-traffic website.

These command-line settings yielded the best compression ratio so far (Linux version of zopfli-krzymod):
zopfli --i1000000 --mb8 --mls4096 --bsr10 --lazy --ohh

(found September 30, 2015)
Description Value Parameter
iterations 1000000  --i1000000
maximum blocks 8  --mb8
maximum length score 4096  --mls4096
block splitting recursion 10  --bsr10
lazy matching in LZ77 yes  --lazy
optimized Huffman headers yes  --ohh
initial random W for iterations 1  --rw1
initial random Z for iterations 2  --rz2

Even Smaller Files Thanks To Defluff

Zopfli's output can be further optimized by the defluff tool.
In this particular case, defluff saves 1 more byte (19986 bytes).

Verify file integrity

After decompression, my uncompressed files are identical to the original ones:

MD5:
curl --silent --compressed http://knockoutjs.com/downloads/knockout-3.3.0.js --location | md5sum
a2d65ff1c1198342386c635ee984d207  -
curl --silent --compressed http://minime.stephan-brumme.com/files/knockout/knockout-3.3.0.min.zopfli.js.gz | md5sum
a2d65ff1c1198342386c635ee984d207  -

SHA1:
curl --silent --compressed http://knockoutjs.com/downloads/knockout-3.3.0.js --location | sha1sum
59d677470345e3b4f28a2ee4af86879de1a0ad5b  -
curl --silent --compressed http://minime.stephan-brumme.com/files/knockout/knockout-3.3.0.min.zopfli.js.gz | sha1sum
59d677470345e3b4f28a2ee4af86879de1a0ad5b  -

All listed CDNs deliver identical contents:
CDN Size (compressed) MD5 (uncompressed) Timestamp
Microsoft 25923 bytes a2d65ff1c1198342386c635ee984d207 (invalid)
Boot 23772 bytes a2d65ff1c1198342386c635ee984d207 (invalid)
cdnjs 20895 bytes a2d65ff1c1198342386c635ee984d207 (invalid)
cdnhttps 20716 bytes a2d65ff1c1198342386c635ee984d207 December 24, 2015 @ 07:33
unpkg 20692 bytes a2d65ff1c1198342386c635ee984d207 July 11, 2016 @ 16:43
jsdelivr 20684 bytes a2d65ff1c1198342386c635ee984d207 (invalid)

Note: only the MD5 hashes are shown to keep things simple.

Other Versions

Available Knockout versions at minime.stephan-brumme.com:

3.4.2, 3.4.1, 3.4.0,
3.3.0,
3.2.0,
3.1.0,
3.0.0,
2.3.0,
2.2.1, 2.2.0,
2.1.0,
2.0.0,
1.2.1, 1.2.0,
1.1.2, 1.1.1, 1.1.0

The project site contains an overview how well these versions were compressed.
Other interesting projects are AngularJS, BackboneJS, Bootstrap, D3, Dojo, Ember, jQuery, lodash, React, Socket.IO, ThreeJS, UnderscoreJS and Vue.

Changelog

Best Zopfli parameters so far:
Size Improvement Parameters Found
19987 bytes -1 byte zopfli --i1000000 --mls4096 --bsr10 --lazy --ohh September 30, 2015 @ 12:34
19988 bytes -1 byte zopfli --i100000 --mls4096 --bsr10 --lazy --ohh September 11, 2015 @ 19:26
19989 bytes -2 bytes zopfli --i10000 --mls4096 --bsr10 --lazy --ohh September 10, 2015 @ 12:12
19991 bytes -1 byte zopfli --i1000 --mls4096 --bsr10 --lazy --ohh September 10, 2015 @ 11:44
19992 bytes -1 byte zopfli --i1000 --mls4096 --bsr12 --lazy --ohh September 10, 2015 @ 11:44
19993 bytes zopfli --i100 --mls4096 --bsr10 --lazy --ohh September 10, 2015 @ 11:37

If there are multiple parameter sets yielding the same compressed size, only the first one found is shown.

Most recent activity on July 11, 2016 @ 16:43.

Heatmaps

This Zopfli heatmap visualizes how compression changes when modifying the --bsr and --mls parameter.
Cell's contents is the best filesize achieved (in bytes, hover with mouse over cells to see number of iterations).

Good parameters are green, bad are red. The best and worst are bold as well.
The brightness of the blue background color indicates how many iterations were processed:
10,000, 100,000 or 1,000,000.
bsr \ mls
2 4 8 16 32 64 128 256 512 1024 2048 4096 8192 16384 32768
bsr \ mls
2 4 8 16 32 64 128 256 512 1024 2048 4096 8192 16384 32768
20003 20008 20006 20003 20003 20003 20002 20003 20002 20002 20000 20000 20012 20012 20008
20010 20012 19998 20000 20001 20002 20003 20002 20002 20009 20002 20010 19999 20008 20003
20011 20007 20010 20010 20002 20005 20003 20002 20000 20005 20003 19999 20000 19999 20008
20004 20013 19997 20013 19998 20000 20014 19998 20000 20001 20003 20001 19995 20002 19992
20010 20010 20010 20003 20003 20012 20002 20002 20000 20002 20002 19993 19999 19998 20006
20010 20008 20002 19998 19999 20010 20003 20001 20000 20004 20003 19993 20009 19997 19997
20010 20012 20003 20002 20003 20002 20001 20002 20003 20002 20003 19987 19999 20000 20006
20010 20010 20000 20000 20003 20012 20003 20002 20002 20002 20002 19994 20010 20000 20014
20010 20012 20012 20015 20014 20014 20008 19999 20001 20005 20003 19987 20002 20000 20009
20010 20008 19998 20013 19998 20013 20006 20005 20003 20002 20003 20000 19993 19997 20007
20012 20012 20014 20013 19997 20013 20003 20003 20002 20001 20003 19994 20007 19999 20014
20007 20007 20010 20002 20003 20010 20003 20005 20002 20001 20003 20000 20010 19999 20003
20010 20012 19997 20014 19997 20005 20013 19993 20004 20002 20003 20001 19995 20000 20000
20007 20007 20000 20003 20003 20015 20003 20002 20003 20002 20002 20000 19999 20000 20009
20007 20012 20002 20015 20015 20005 20002 20001 20002 20002 20003 20009 20009 20002 20008
20010 20010 20011 19998 20002 20005 20004 20002 20000 20002 20003 20006 20008 19999 20014
20010 20001 19997 19988 20002 20005 20004 20005 20001 20001 20003 20008 19999 19999 19997
20009 20012 20011 19999 20002 20005 20003 20002 20003 20006 20003 20008 19999 19999 20009
20010 20008 20003 20003 20003 20005 20003 20002 20000 20001 20002 20000 20011 19999 20008
20009 20010 20000 20002 20014 20005 20002 20002 20001 20002 20003 20000 20011 19997 20010
20010 20008 20010 20001 20013 20006 20006 19999 20003 20005 20003 19987 20001 20002 20005
20010 20008 19999 19999 20002 20015 20003 20002 20003 20004 20003 20000 20001 20000 20008
20010 20000 20003 20003 20000 20013 20002 20002 20003 20004 20003 19987 20008 20004 19999

Due to the Monte Carlo design of my search algorithm, not all parameters have reached the same number of iterations yet:
Iterations Min. Bytes Reduction Coverage
100 19993 bytes 100%
1,000 19991 bytes -2 bytes 100%
10,000 19989 bytes -2 bytes 100%
100,000 19988 bytes -1 byte 9.28%
1,000,000 19987 bytes -1 byte 4.06%
10,000,000

KZIP has far less options available for tuning/optimization. I only played around with the number of blocks (parameter -n):
Blocks Min. Bytes Compared To Best Zopfli Compared To Best KZIP
20085 bytes +98 bytes (+0.49%) +49 bytes
20085 bytes +98 bytes (+0.49%) +49 bytes
20036 bytes +49 bytes (+0.25%)
20053 bytes +66 bytes (+0.33%) +17 bytes
20096 bytes +109 bytes (+0.55%) +60 bytes
20092 bytes +105 bytes (+0.53%) +56 bytes
20131 bytes +144 bytes (+0.72%) +95 bytes
20170 bytes +183 bytes (+0.92%) +134 bytes
20196 bytes +209 bytes (+1.05%) +160 bytes

Non-DEFLATE Algorithms

Archivers based on completely different compression algorithms often produce superior results.
Unfortunately, browsers only support gzip compression at the moment.
Algorithm Program Parameters Size Compared To Best Zopfli
RAR (proprietary) RAR rar a -m5 -md64m -mc63:128t -mt1 16940 bytes -3047 bytes (-15.24%)
ZPAQ (Wikipedia) zpaq zpaq -method 69 17087 bytes -2900 bytes (-14.51%)
PPMd (Wikipedia) 7zip 7za a -mx=9 -m0=ppmd 18222 bytes -1765 bytes (-8.83%)
Brotli (Wikipedia) brotli brotli -q 11 18358 bytes -1629 bytes (-8.15%)
Burrows-Wheeler transform (Wikipedia) bzip2 bzip2 -9 18862 bytes -1125 bytes (-5.63%)
LZMA2 (Wikipedia) xz xz -9 19568 bytes -419 bytes (-2.10%)
ZSTD (Wikipedia) zstd zstd -19 19947 bytes -40 bytes (-0.20%)

Detailled Analysis

I wrote a DEFLATE decoder in Javascript. Click the button below to start a client-side analysis of the smallest gzipped files (may take a second):


Notes: pigz is a fast open source multi-threaded implementation of gzip written by one of the original authors of gzip.
However, when using compression level 11, pigz actually switches to the slower Zopfli algorithm and isn't multi-threaded anymore.
KrzyMOD's extensions to Zopfli offer the highest level of configuration and is therefore used for my brute-force search.
Ken Silverman wrote the closed-source KZIP compression program and Jonathon Fowler ported it to Linux.
Defluff was created by Joachim Henke; DeflOpt is a tool by Ben Jos Walbeehm.

website made by Stephan Brumme in 2015 and still improving in 2017.
all timestamps are displayed in central european time. see my changelog.
no flash, not even images or external css files - and everything squeezed into a single html file.
which was handsomely compressed before releasing it into the wild internet - obviously.

please visit my homepage and my blog, too.
email: minime (at) stephan-brumme.com