How to Install Anaconda on Mac: Python and Jupyter Notebook Setup

To install Anaconda on Mac, download the Apple Silicon graphical installer from Anaconda's download page, run the .pkg, and reopen Terminal so that (base) appears in your prompt. The install itself takes minutes. What trips people up comes afterwards.

Three problems account for most failed setups: an Intel Mac that no longer gets new Anaconda packages, a conda: command not found error caused by the zsh shell, and a notebook that cannot see a package you just installed. This guide is for data science and Python learners on macOS, and it covers the fix for each, plus leaving Apple's own Python alone.

This guide follows Anaconda's macOS installation instructions as of September 2026, when the latest release was Anaconda Distribution 2026.07. On Windows, follow the Windows version of this guide instead.

Before you install Anaconda on Mac

Check these first:

• Your chip. Open the Apple menu, choose About This Mac, and look for "Apple M1" (or later) or "Intel". In Terminal, uname -m prints arm64 on Apple Silicon.
• macOS version. Anaconda's system requirements list 64-bit macOS 12.1 or later on Apple Silicon.
• Disk space. Anaconda's comparison of Anaconda and Miniconda gives about 9.7 GB of install space for the full distribution and about 900 MB for Miniconda.
• Licence terms. Under Anaconda's terms of service, use is free for individuals, eligible academic and non-profit institutions, and for-profit organisations with 200 or fewer employees and contractors. Larger for-profit organisations need a paid business licence, so check with IT first if that applies to you.

What about Intel Macs?

Anaconda's documentation states that it stopped building packages for Intel Macs (osx-64) on 15 August 2025. Older Intel installers remain in Anaconda's archive, but they will not receive new packages. If you are on an Intel Mac, therefore, expect to fall behind on package versions, and consider moving to a newer machine or to another Python setup for long-term work.

Should you use the Python that comes with macOS?

No. Leave it for Apple's tools and do your own work in Anaconda. macOS has a /usr/bin/python3. It belongs to Apple's Command Line Tools, and on a Mac without those tools it only offers to install them.

Never sudo pip install into it. Doing so can break the tools that rely on it, and a software update can replace the whole thing, taking your packages with it.

Anaconda sidesteps this by installing into its own folder.

How do you install Anaconda on Mac step by step?

1. Download. Go to anaconda.com/download and choose the 64-Bit (Apple silicon) Graphical Installer. Registration is optional; the page offers a "Skip registration" link.
2. Open the .pkg. Double-click it, read the Read Me, and select Continue.
3. Agree to the terms of service.
4. Choose the destination. Anaconda recommends Install for all users of this computer, which installs into /opt/anaconda3. Alternatively, Install on a specific disk lets you pick another location.
5. Install, entering your Mac password if prompted, then close the installer.
6. Open a new Terminal window. You should see (base) at the start of the prompt.

If you prefer the command line, Anaconda's terminal installer guide uses the .sh installer instead. At the time of writing, the file was named for version 2026.07-1:

bash ~/Anaconda3-2026.07-1-MacOSX-arm64.sh

That route installs into ~/anaconda3 by default. Accept the licence, confirm the location, and answer yes when it offers to initialise conda.

Why is conda not found after installing on Mac?

Usually because your current Terminal session started before conda edited the zsh configuration. Reopen Terminal or reload ~/.zshrc. macOS has used zsh as the default shell since Catalina, so conda writes to ~/.zshrc, not ~/.bash_profile. Older tutorials often get this wrong.

Check which shell you are in:

echo $SHELL

If conda is not found after installing, either restart Terminal or run:

source ~/.zshrc

Still missing? Initialise explicitly with the full path for your install:

/opt/anaconda3/bin/conda init zsh      # graphical installer
~/anaconda3/bin/conda init zsh         # command-line installer

Then reopen Terminal.

If you dislike having base auto-activate in every shell, turn it off:

conda config --set auto_activate false

Older conda releases call this setting auto_activate_base. Current conda accepts that name as an alias and prints a warning that it is setting auto_activate instead.

How do you check Anaconda installed correctly?

conda --version
python --version
which python        # should point inside /opt/anaconda3 or ~/anaconda3

That which check is the useful one, because it confirms you are running Anaconda's Python rather than Apple's. You can also run conda list to see the installed packages.

How do you create a conda environment for a project?

Run conda create --name <project> python=3.12, activate it, and install packages there rather than into base. One environment per project keeps version conflicts contained:

conda create --name myproject python=3.12
conda activate myproject
conda install pandas numpy matplotlib scikit-learn jupyter

The first time you use Anaconda's default channels, conda may ask you to accept their terms of service. Answer a, or run conda tos accept once, as Anaconda's ToS command-line guide describes.

Management commands:

conda env list
conda list                          # packages in the active environment
conda deactivate
conda env remove --name myproject

Sharing:

conda env export --from-history > environment.yml
conda env create -f environment.yml

--from-history records only the packages you explicitly requested, which makes the file portable across platforms. A plain export, by contrast, pins macOS-specific builds and will fail on a colleague's Windows machine.

How do you use a conda environment in Jupyter?

Install ipykernel in the environment and register it as a kernel, because environments do not appear in Jupyter automatically:

conda activate myproject
conda install ipykernel
python -m ipykernel install --user --name myproject --display-name "Python (myproject)"

Skipping this is the most common reason a notebook reports a package as missing when you have just installed it.

Launch Jupyter

Activate the environment, move to the project folder and start JupyterLab:

conda activate myproject
cd ~/projects/analysis
jupyter lab

Jupyter roots its file browser wherever you launched it, so start from the project folder. Both jupyter lab and jupyter notebook are covered on jupyter.org's install page.

Stop the server with Ctrl+C twice. Closing the browser tab leaves the server running.

Do data science packages work on Apple Silicon?

Yes, for the common ones. Most scientific packages ship native arm64 builds. For anything missing from Anaconda's channels, conda-forge is the first place to look:

conda install -c conda-forge package-name

For machine learning, Apple Silicon GPUs are available to PyTorch through its MPS backend and to TensorFlow through Apple's tensorflow-metal plugin. Follow each project's current instructions, because the package names have changed before: TensorFlow 2.12 and earlier used tensorflow-macos.

Quick reference: after you install Anaconda on Mac

Common problems on macOS

zsh: command not found: conda The shell configuration has not been applied. Restart Terminal, or run conda init zsh with the full path shown above.

which python shows /usr/bin/python3 No conda environment is active, or the PATH change did not take. Run conda activate myproject and check again.

Slow dependency solving Since conda 23.10 the faster libmamba solver has been the default, according to the conda-libmamba-solver guide. If yours is older, update conda with conda update -n base conda.

Jupyter cannot find an installed package You are on the wrong kernel. Check inside the notebook:

import sys
print(sys.executable)

If that path is not inside your environment, select the correct kernel from the Jupyter menu.

Should you install Miniconda instead?

Choose Miniconda if disk space matters or you prefer to install only what you use. Miniconda gives the same conda tooling with only Python and its dependencies, so it needs about 900 MB rather than about 9.7 GB. Install it from the same download page, then add only what you need:

conda create -n ds python=3.12 pandas numpy matplotlib scikit-learn jupyter

Alternatively, Homebrew plus venv is a perfectly good non-conda route, and preferable if your work is general Python rather than data science.

Related reading

The Windows version of this guide covers the PATH checkbox and Anaconda Prompt. Once set up, see Python's building blocks, the NumPy tutorial series and Git and GitHub.