conda Cheat Sheet

conda, Anaconda, and Miniconda

• conda: The package and environment manager
• Miniconda: A minimal installer with only conda, Python, and their dependencies
• Anaconda Distribution: A full installer with hundreds of data science packages and Anaconda Navigator
• Miniforge: A community installer preset to conda-forge
• Licensing: Anaconda’s repository has Terms of Service, and large organizations may need a paid license

Look these up in the Real Python reference:

• Conda
• Anaconda
• Anaconda Navigator

Not sure which installer to pick?

Help me choose a conda installer

Installation and Setup

• Install Miniconda without admin rights in a user-writable folder.
• Miniforge installs conda preset to conda-forge, with no Anaconda Terms of Service to accept.
• Answer yes when the installer asks to initialize conda, then restart your shell.
• On Windows, use the Anaconda Prompt or conda init powershell.

$ U=https://repo.anaconda.com/miniconda
$ F=Miniconda3-latest-Linux-x86_64.sh
# Apple Silicon Macs use this file instead:
# F=Miniconda3-latest-MacOSX-arm64.sh
$ curl -O "$U/$F"
$ bash "$F"
$ source ~/.bashrc   # Or ~/.zshrc

PS> $exe = "Miniconda3-latest-" +
>>   "Windows-x86_64.exe"
PS> Invoke-WebRequest -OutFile $exe -Uri `
>> "https://repo.anaconda.com/miniconda/$exe"
PS> Start-Process $exe   # Run the installer

$ conda --version
conda 26.9.3
$ conda info  # Base environment and channels
$ conda update -n base conda

$ conda init bash   # Or zsh or powershell
$ conda config --set auto_activate false

Want to go deeper on your setup?

Show me where to learn more about Python setups

Environments

• Create one environment per project.
• Never install project packages into base.
• Look for * next to the active environment in conda env list.
• Give each environment the Python version it needs.

$ conda create -n ds python=3.13
$ conda activate ds
(ds) $ python --version
Python 3.13.16
(ds) $ conda deactivate

$ conda env list
base     /home/me/miniconda3
ds       /home/me/miniconda3/envs/ds
$ conda create -n ds2 --clone ds
$ conda rename -n ds2 ds-old
$ conda remove -n ds-old --all

$ conda run -n ds python script.py
$ conda create -p ./env python=3.13
$ conda activate ./env   # Activate by path

Think you’ve got environments down?

Quiz me on conda environments

Packages

• Rely on conda to resolve dependencies, including non-Python libraries.
• Add -n ENV to target an environment that isn’t active.
• Add -y to skip the confirmation prompt.

$ conda search pandas
$ conda search "*keras*"   # Wildcard
$ conda install numpy pandas
$ conda install numpy=2.2
$ conda install "numpy>=2,<3"

$ conda list               # All packages
$ conda list numpy
# Name  Version  Build           Channel
numpy   2.5.4    py313hb5f73ae_0 conda-forge
$ conda update numpy
$ conda update --all
$ conda remove numpy

Analyzing data next? Pair this sheet with the pandas Cheat Sheet.

Wondering what an update touches?

Show me what conda update –all can change

Channels

• defaults is Anaconda’s repository, and conda-forge is community-run.
• Mixing channels in one environment can cause conflicts, so pick one main channel.
• Miniconda ships with defaults, so remove it to use only conda-forge.
• Strict priority ignores lower-priority channels for any package that a higher-priority channel has.

$ conda install -c conda-forge scipy
$ conda install conda-forge::scipy

$ conda config --system \
    --remove channels defaults
$ conda config --add channels conda-forge
$ conda config --append channels bioconda
$ conda config --set channel_priority strict
$ conda config --show channels
channels:
  - conda-forge
  - bioconda

Channel priority still fuzzy?

Explain channel priority another way

Environment Files

• --from-history keeps only what you asked for, so the file works across platforms, but it leaves out pip packages. Add those to pip: by hand.
• --prune removes conda packages that are no longer in the file, but pip packages stay.

name: myproject
channels:
  - conda-forge
dependencies:
  - python=3.13
  - numpy
  - pip
  - pip:
      - rich

$ conda env export --from-history \
    > environment.yml
$ conda env create -f environment.yml
$ conda env update -f environment.yml \
    --prune

Coming from requirements.txt?

Help me compare environment.yml and requirements.txt

Packages From PyPI With pip

• Install everything you can with conda first, then use pip.
• Install pip into the environment and never use pip --user.
• After using pip, recreate the environment if you need more conda packages.
• List pip packages under pip: yourself, since --from-history exports skip them.
• Install pure-Python PyPI packages with conda 26.9 and later through the conda-pypi channel.

$ conda activate myproject
(myproject) $ conda install pip
(myproject) $ python -m pip install unipath
(myproject) $ conda list unipath
unipath   1.1   pypi_0   pypi

Mixing conda and pip?

Show me what breaks if I run conda after pip

History, Cleanup, and Configuration

• Most conda package changes create a revision, and revision 0 is the freshly created environment.
• Revisions and rollbacks don’t track pip packages.
• conda clean removes unused packages and cached tarballs.
• User settings live in ~/.condarc.

$ conda list --revisions
2026-10-15 08:40:44  (rev 0)
    +_openmp_mutex-4.5 (conda-forge/linux-64)
    ...
$ conda install --revision 0

$ conda clean --all --dry-run
$ conda clean --all

$ conda config --show-sources
$ conda config --show channel_priority
$ conda COMMAND --help

Ready to test yourself?

Quiz me on conda housekeeping

Ready to go beyond the cheat sheet?

Build me a conda learning plan

You can download this information as a printable cheat sheet:

Free Bonus: conda Cheat Sheet

Get a conda Cheat Sheet (PDF) with the commands for environments, packages, channels, and environment files on one printable page:

Send My conda Cheat Sheet »