Operating System
The ArcGIS API for Python (2.4.0 and greater releases) is compatible with 64-bit versions of Windows, macOS (M-Series and Intel), and Linux.
Python Version
Python 3.10.x to 3.12.x is required to use the ArcGIS API for Python 2.3.1 and greater releases.
Dependencies
The full power of the ArcGIS API for Python is best experienced when all its dependencies are installed. However, specific tasks such as GIS administration and content management can be accomplished with a subset of dependencies installed. See Install with minimum Dependencies to install the arcgis package in this manner.
It is recommended to install the arcgis package the default way of either conda install -c esri arcgis arcgis-mapping or pip install arcgis arcgis-mapping. When installed in this manner, all the below dependencies are automatically installed.
Note Most of these packages have dependencies of their own. For a full list of packages installed:
- conda environment, type
conda list -n <environment_name>.- pip environment, type
pip list.
- anywidget
- cachetools
- certifi
- cryptography
- dask >=2024.12.1,<2025.1.0
- geomet
- ipywidgets >=8
- jupyterlab >=4.0.7,<5
- keyring >=23.3.0
- lxml
- matplotlib
- matplotlib-inline
- networkx
- numpy >=1.21.6,<2
- pandas >=2.0.0,<2.3.0
- pillow
- puremagic >=1.15,<2
- pyarrow >=14,<17
- pydantic >=2.8.2,<3
- pylerc
- pyspnego >=0.8.0
- requests >=2.32.3,<3
- requests-gssapi
- requests-kerberos
- requests-oauthlib
- requests-toolbelt
- truststore
- ujson >=3
- urllib3 >=2.1.0,<3
- websocket-client >=1.2.3,<2
- Windows only:
Note: The dependencies above result from installing using
conda. Versions may vary ifarcgisis installed usingpip.
Note: if
arcpyis found in the current python environment, it may be used in various locations. Otherwise,gdalorpyshpwill be used. See Spatially Enabled DataFrame for more information.
Optional Dependencies
There are some other python packages that may be required to use certain functionality in the API, but are not automatically installed. To use that functionality, you must manually call conda install {package_name} or pip install {package_name} for such optional packages.