This section gives more details about the installation process of Bokeh, for those who were unable to complete the process detailed in the Quickstart, or who want more information about the process.


Bokeh is officially supported (and continuously tested) on CPython versions 2.7 and 3.5+ only. Other Python versions may function, possibly in limited capacity. In particular, converting NumPy arrays to lists may be useful with other versions. However, this guidance is only provided as-is, in case it happens to be useful, and does not imply any level of official support for other Python versions. All issues opened related to unsupported Python versions will be closed as invalid.

For basic usage, have the following libraries installed:

  • NumPy
  • Jinja2
  • Six
  • Requests
  • Tornado >= 4.0
  • PyYaml
  • DateUtil

To use the Bokeh server with python 2.7, you also need to install Futures package.

Because the Bokeh client library is mostly concerned with providing a nice Python interface for generating JSON objects which are then consumed by the BokehJS library running in the browser, there shouldn’t be a hard dependency on any of the standard NumPy/SciPy stack. It is entirely possible to use Bokeh with plain Python lists of values. However, the Bokeh plot server does make direct use of NumPy, and it is required to be installed for Bokeh apps to function. Additionally nodejs is required to allow compilation of custom bokeh extensions.

Package Installs

These Bokeh dependencies are best obtained via the Anaconda Python Distribution, which was designed to include robust versions of popular libraries for the Python scientific and data analysis stacks.

If you are already an Anaconda user, you can simply run the command:

conda install bokeh

This will install the most recent published Bokeh release from the Anaconda, Inc. package repository, along with all dependencies.

Alternatively, it is possible to install from PyPI using pip:

pip install bokeh

Sample Data

Some of the Bokeh examples rely on sample data that is not included in the Bokeh GitHub repository or released packages, due to their size. Once Bokeh is installed, the sample data can be obtained by executing the following command at a Bash or Windows prompt:

bokeh sampledata

Alternatively, the following statements can be executed in a Python interpreter:

>>> import bokeh.sampledata

Finally, the location that the sample data is stored can be configured. By default, data is downloaded and stored to a directory $HOME/.bokeh/data. (The directory is created if it does not already exist.) Bokeh looks for a YAML configuration file at $HOME/.bokeh/config. The YAML key sampledata_dir can be set to the absolute path of a directory where the data should be stored. For instance adding the following line to the config file:

sampledata_dir: /tmp/bokeh_data

will cause the sample data to be stored in /tmp/bokeh_data.

Installing from Source

Installing Bokeh from source requires rebuilding the BokehJS library from its CoffeeScript sources. Some additional toolchain support is required. Please consult the Getting Set Up section of the Developer Guide for detailed instructions.

Developer Builds

And easier way to obtain the most recent Bokeh updates without having to worry about building Bokeh yourself is to install a developer build. We typically try to make a new developer build available at least once a week, and sometimes more often.

These builds are being made available on If you are using Anaconda, you can install with conda by issuing the command from a Bash or Windows command prompt:

conda install -c bokeh/channel/dev bokeh

Alternatively you can install with pip from a Bash or Windows command prompt:

pip install --pre -i bokeh --extra-index-url

We attempt to make sure the developer builds are relatively stable, however please be aware they they are not tested as rigorously as standard releases. Any problems or issues reported on the GitHub issue tracker are appreciated.

BokehJS Standalone

If you would like to use BokehJS as a standalone JavaScript library, there are two easy ways to get any published release.

First, released versions of BokehJS is available for download from CDN at, under the following naming scheme:

for the BokehJS JavaScript files, and:

for the BokehJS CSS files.


The CSS must be loaded before the JavaScript library.

The "-widgets" files are only necessary if you are using any of the widgets built into Bokeh in bokeh.models.widgets in your documents.

As a concrete example, the links for version 0.12 are:



For releases 0.12.2 and after, the BokehJS API has been branched to a separate file. It is also available for download from CDN at under the name bokeh-api using the above naming scheme. It must be loaded after the JavaScript library.