How do I check which version of NumPy I'm using?

Question

How can I check which version of NumPy I'm using?

probable you just need to install specific version of [numpy](https://stackoverflow.com/questions/1520234/how-do-i-check-which-version-of-numpy-im-using/53898417#53898417) — prosti, Dec 22 '18 at 18:49

score 452 · Accepted Answer · edited Apr 22 '16 at 14:21

452

import numpy
numpy.version.version

edited Apr 22 '16 at 14:21

David Stansby

1,491
13
17

answered Oct 05 '09 at 14:02

SilentGhost

287,765
61
300
288

2

@Dominic Rodger: yeah, but your is more general to any module that cares to set a `__version__`. – Esteban Küber Oct 05 '09 at 14:13
68

This is not the public API, numpy.__version__ is. – David Cournapeau Nov 22 '11 at 10:10
13

Actually `import numpy ; numpy.version.version` . The lack of `import numpy` through me, an obvious newbie. – mmorris Apr 20 '12 at 01:20
13

Since the use of `__version__` in recommended in PEP8 and most packages support `__version__` vs the non standard `version.version` I think that this answer should be treated more as a curiosity than an accepted method. Use `numpy.__version__` or `.__version__` as [Dominic Rodger's answer recommends](http://stackoverflow.com/a/1520264/298607) Parse the version (and create your own version strings) as recommended in PEP 386 / PEP 440. – dawg Apr 02 '14 at 16:09
Actually, in NumPy 1.20, `numpy.version.version` is typed, and `numpy.__version__` is not (could be an oversight). – Henry Schreiner Jan 31 '21 at 16:43

score 265 · Answer 2 · answered Oct 05 '09 at 14:01

265

>> import numpy
>> print numpy.__version__

answered Oct 05 '09 at 14:01

Dominic Rodger

94,357
33
195
210

63

This is the API we numpy developers will support. numpy.version.version is an implementation detail that should not be relied upon. – Robert Kern Oct 06 '09 at 00:18
well, this works for most well-written python packages. – fast tooth Oct 23 '14 at 03:02
1

This is the PEP8 Way(tm) +1 – Burhan Khalid Sep 27 '16 at 09:01

score 72 · Answer 3 · edited Jun 27 '18 at 20:25

72

From the command line, you can simply issue:

python -c "import numpy; print(numpy.version.version)"

Or:

python -c "import numpy; print(numpy.__version__)"

edited Jun 27 '18 at 20:25

Peter Mortensen

30,030
21
100
124

answered Dec 24 '09 at 12:10

meduz

3,528
1
24
38

1

This is actually very nice as it allows you to check the version of numpy even if you have two different versions running on two different versions of python. py -2 -c "import numpy; print(numpy.version.version)" py -3 -c "import numpy; print(numpy.version.version)" – JSWilson Aug 18 '20 at 20:57

score 34 · Answer 4 · answered May 09 '17 at 05:59

34

Run:

pip list

Should generate a list of packages. Scroll through to numpy.

...
nbpresent (3.0.2)
networkx (1.11)
nltk (3.2.2)
nose (1.3.7)
notebook (5.0.0)
numba (0.32.0+0.g139e4c6.dirty)
numexpr (2.6.2)
numpy (1.11.3) <--
numpydoc (0.6.0)
odo (0.5.0)
openpyxl (2.4.1)
pandas (0.20.1)
pandocfilters (1.4.1)
....

answered May 09 '17 at 05:59

Lavanya Shukla

341
3
2

1

Or `pip freeze` if in a virtual environment? – Hassan Baig May 09 '17 at 06:28

score 18 · Answer 5 · answered Jun 03 '13 at 19:40

18

You can also check if your version is using MKL with:

import numpy
numpy.show_config()

answered Jun 03 '13 at 19:40

David C

6,789
4
48
65

score 11 · Answer 6 · edited Jun 27 '18 at 20:26

11

We can use pip freeze to get any Python package version without opening the Python shell.

pip freeze | grep 'numpy'

edited Jun 27 '18 at 20:26

Peter Mortensen

30,030
21
100
124

answered Nov 02 '15 at 02:06

Ajay Gupta

3,142
1
21
30

2

That only works if you installed numpy via pip, not via brew or apt-get, for example. – Rafael Almeida Aug 02 '16 at 20:49

score 11 · Answer 7 · answered Sep 20 '17 at 19:47

11

You can try this:

pip show numpy

answered Sep 20 '17 at 19:47

SDt

362
1
3
10

score 8 · Answer 8 · edited Jun 20 '20 at 09:12

If you're using NumPy from the Anaconda distribution, then you can just do:

$ conda list | grep numpy
numpy     1.11.3     py35_0

This gives the Python version as well.

If you want something fancy, then use `numexpr`

It gives lot of information as you can see below:

In [692]: import numexpr

In [693]: numexpr.print_versions()
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Numexpr version:   2.6.2
NumPy version:     1.13.3
Python version:    3.6.3 |Anaconda custom (64-bit)|
                   (default, Oct 13 2017, 12:02:49)
[GCC 7.2.0]
Platform:          linux-x86_64
AMD/Intel CPU?     True
VML available?     False
Number of threads used by default: 8 (out of 48 detected cores)
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

score 5 · Answer 9 · answered Feb 22 '19 at 20:10

5

You can get numpy version using Terminal or a Python code.

In a Terminal (bash) using Ubuntu:

pip list | grep numpy

In python 3.6.7, this code shows the numpy version:

import numpy
print (numpy.version.version)

If you insert this code in the file shownumpy.py, you can compile it:

python shownumpy.py

or

python3 shownumpy.py

I've got this output:

1.16.1

answered Feb 22 '19 at 20:10

Rogelio Prieto

359
3
10

Just a slight caution that it's possible that you may have python and python 3 both installed with numpy. Then when doing the `pip list | grep numpy` method it will show one of the two (typically the python 3's numpy version). When you run the `shownumpy.py` program on both python and python 3, they will show you exactly what version is on each respective python environment. – Caleb Feb 22 '19 at 20:37

score 3 · Answer 10 · edited Jan 14 '19 at 13:59

3

For Python 3.X print syntax:

python -c "import numpy; print (numpy.version.version)"

Or

python -c "import numpy; print(numpy.__version__)"

edited Jan 14 '19 at 13:59

Guillaume Jacquenot

10,118
5
41
48

answered Nov 03 '15 at 01:51

user1729077

39
2

1

Exact duplicate of the answer of @meduz . For python 3, it is `print(numpy.__version__)`, not `print numpy.__version__` – francis Mar 23 '16 at 15:02

score 3 · Answer 11 · answered Jul 09 '17 at 11:34

3

import numpy
print numpy.__version__

answered Jul 09 '17 at 11:34

Dinusha Thilakarathna

412
6
13

Tanner Dolby · Answer 12 · 2019-12-03T01:51:37.077

3

Just a slight solution change for checking the version of numpy with Python,

import numpy as np 
print("Numpy Version:",np.__version__)

Or,

import numpy as np
print("Numpy Version:",np.version.version)

My projects in PyCharm are currently running version

1.17.4

edited Dec 03 '19 at 01:51

answered Dec 03 '19 at 01:35

Tanner Dolby

3,652
3
9
19

score 3 · Answer 13 · answered Nov 01 '20 at 18:53

3

Simply

pip show numpy

and for pip3

pip3 show numpy

Works on both windows and linux. Should work on mac too if you are using pip.

answered Nov 01 '20 at 18:53

Sajidur Rahman

2,314
23
25

score 2 · Answer 14 · edited Jun 27 '18 at 20:28

2

In a Python shell:

>>> help()
help> numpy

edited Jun 27 '18 at 20:28

Peter Mortensen

30,030
21
100
124

answered Jul 26 '17 at 09:28

LMB

232
1
3
15

score 0 · Answer 15 · answered Jul 19 '18 at 15:00

0

Pure Python line that can be executed from the terminal (both 2.X and 3.X versions):

python -c "import numpy; print(numpy.version.version)"

If you are already inside Python, then:

import numpy
print(numpy.version.version)

answered Jul 19 '18 at 15:00

Ginés Hidalgo

597
6
19

score 0 · Answer 16 · answered Dec 22 '18 at 18:47

It is good to know the version of numpy you run, but strictly speaking if you just need to have specific version on your system you can write like this:

pip install numpy==1.14.3 and this will install the version you need and uninstall other versions of numpy.

score 0 · Answer 17 · answered Aug 19 '21 at 15:05

0

For Windows

pip list | FINDSTR numpy

For Linux

pip list | grep numpy

answered Aug 19 '21 at 15:05

Arihant Jain

721
3
11

How do I check which version of NumPy I'm using?

17 Answers17

If you want something fancy, then use `numexpr`

Linked

How do I check which version of NumPy I'm using?

17 Answers17

If you want something fancy, then use numexpr

Linked

If you want something fancy, then use `numexpr`