Install Anaconda on macOS (2025): Apple Silicon, conda init, and Common Fixes

Installing Anaconda on a Mac used to be straightforward: download, click, done. Today, architecture differences (Apple Silicon vs. legacy Intel), shell initialization, and PATH conflicts make it easy to end up with “conda: command not found,” a slow Intel build running under Rosetta, or a Jupyter launch that never starts.

I avoid those failure modes by following a simple checklist: pick the correct installer, verify the checksum, let the installer run conda init, restart Terminal, and test with a fresh environment. The steps below reflect what works reliably on current macOS versions (Sonoma 14 and Sequoia 15).

What Is Anaconda?

Anaconda is a Python distribution that bundles Python and hundreds of scientific packages (NumPy, pandas, SciPy, scikit-learn, and more). It includes the conda package and environment manager so you can install prebuilt packages and keep project dependencies isolated.

Important Licensing Note (2025 Update)

As of July 2025, the free Anaconda Individual Edition supports use by individuals, educational institutions, non-profits, and for-profit organizations with 200 or fewer employees or contractors. Organizations with more than 200 employees or contractors must obtain a paid Business (or Enterprise) license under Anaconda’s terms.

See the current Anaconda Terms of Service for details.

Choose the Right Installer for Your Mac

Before downloading, confirm your Mac’s architecture and pick the matching installer from the official site.

• Apple Silicon (M1/M2/M3): Use the macOS ARM64 build (osx-arm64). This runs natively and performs best.
• Intel Macs: New Intel (osx-64) builds are largely discontinued as of mid-2025. You can use archived installers or consider Miniforge/Miniconda instead.

Check your architecture:

uname -m # prints 'arm64' on Apple Silicon, 'x86_64' on Intel

Download the installer from the official Anaconda site: https://www.anaconda.com/download

Graphical Install (.pkg) on macOS

The graphical installer is the quickest way to install Anaconda on macOS and is suitable for most users.

Verify the installer checksum

shasum -a 256 ~/Downloads/Anaconda3-2025.06-0-MacOSX-arm64.pkg

If the checksum doesn’t match, re-download or remove macOS’s quarantine attribute before rechecking:

xattr -dr com.apple.quarantine ~/Downloads/Anaconda3-2025.06-0-MacOSX-arm64.pkg

Run the installer

Double-click the .pkg file. If prompted, provide admin credentials.

Choose an install location:

• Default (no sudo): /Users/<username>/anaconda3

• System-wide (requires sudo): /opt/anaconda3

Allow the installer to run conda init when prompted. Close the installer, then close and reopen Terminal.

Verify installation

which conda conda list | head python3 --version

If everything worked, which conda should point to your Anaconda folder and conda list should print many packages.

Command-Line Install (.sh) on macOS

The command-line installer gives you full control over the destination and is suitable for scripting or headless setups.

Verify checksum

shasum -a 256 ~/Downloads/Anaconda3-2025.06-0-MacOSX-arm64.sh

Run the installer

cd ~/Downloads bash Anaconda3-2025.06-0-MacOSX-arm64.sh

When asked about the install path:

• /opt/anaconda3 for a system-wide install (admin rights required), or

• ~/anaconda3 for a user-local install (no admin required).

Answer “yes” to conda init, then close and reopen Terminal.

Verify installation

which conda conda info 

Modern macOS uses zsh by default, so your profile file is typically ~/.zshrc.

If conda is missing after install, initialize it manually:

echo $SHELL # should print /bin/zsh conda init zsh # Then close and reopen Terminal which conda

If you use Bash:

conda init bash

Do not edit or remove the system Python in /usr/bin.

Verify Your Installation

Quick checks to confirm that Anaconda and Python are working:

conda list | head python3 --version which python3 python3 -c "import sys; print(sys.executable)"

Optional: confirm Jupyter runs:

jupyter lab # or: jupyter notebook

Create and Use Isolated Environments

Avoid working in the base environment. Create one per project.

# Create a new environment
conda create -n analytics-env python=3.11 -y

# Activate the environment
conda activate analytics-env

# Install common libraries
conda install numpy pandas -y

# Verify installation
python3 -c "import numpy as np, pandas as pd; print(np.__version__, pd.__version__)"

Deactivate when done:

conda deactivate

Prefer Conda for Packages (Use pip as Needed)

Install scientific packages with conda when possible. Use pip only when unavailable via conda.

# conda-first approach
conda install scikit-learn

# fallback to pip (if the package isn't available via conda)
pip install package-name

Alternatives: Miniconda, Miniforge, and Mambaforge

If you want a smaller footprint or need a license-compliant option for enterprise use, consider these:

• Miniconda: Minimal installer (~400 MB) with just Python and conda.
• Miniforge / Mambaforge: Community installers defaulting to conda-forge (fully open-source, no commercial restriction).
• Mambaforge adds mamba for much faster dependency resolution.

Installers: https://github.com/conda-forge/miniforge

Common Issues and Fixes

conda: command not found

conda init zsh 
# close and reopen Terminal which conda

Wrong architecture on Apple Silicon

# Check system architecture (hardware)
uname -m

# Check Python interpreter architecture
python3 -c "import platform; print(platform.machine())"

If output is x86_64, uninstall and reinstall the ARM64 build.

Intel Mac support ending

As of mid-2025, new Intel (osx-64) builds are no longer maintained. Use archived installers or Miniforge/Miniconda.

Installer cannot write to /opt

Re-run the installer and change the destination to your home directory (e.g., ~/anaconda3).

Homebrew installations

Avoid installing Anaconda via Homebrew — it’s deprecated and often breaks PATH resolution. Use official or community installers.

Duplicate Anaconda paths

Edit your shell profile and remove duplicates:

# Open your shell profile for editing
nano ~/.zshrc    # or: nano ~/.bash_profile

# (Inside the file) remove extra PATH entries for older Anaconda installs
# Then save and exit (Ctrl+O, Enter, Ctrl+X)

# Reload your updated profile
source ~/.zshrc

# Verify that conda is now pointing to the correct path
which conda

Jupyter fails to open

Ensure the environment is active and jupyterlab is installed there:

# Activate your environment
conda activate analytics-env

# Install JupyterLab inside the environment
conda install jupyterlab -y

# Launch JupyterLab
jupyter lab

Don’t remove system Python

macOS includes a system Python. Leave it untouched. Always use the Python from your conda environment.

Uninstall or Update Anaconda

Update safely:

# Update conda itself
conda update conda

# (Optional) Update all packages in the current environment — use cautiously
conda update --all

Clean and remove:

# Install the Anaconda cleanup utility
conda install anaconda-clean -y

# Run the cleanup tool (removes configs, cache, and settings)
anaconda-clean --yes

# Manually delete the Anaconda installation directory
# (Use caution — this permanently removes the files)
rm -rf ~/anaconda3   # or: sudo rm -rf /opt/anaconda3

Also remove initialization lines from ~/.zshrc or ~/.bash_profile.

Conclusion

Successful Anaconda installs on macOS come down to a few critical choices:

• Use the correct architecture (ARM64 on Apple Silicon).
• Verify your download.
• Allow conda init and restart Terminal.
• Work in project-specific environments.
• Respect licensing: use Miniforge or Miniconda for enterprise or commercial use.

With these steps, conda, Python, and Jupyter will behave predictably, and you’ll avoid both technical and licensing pitfalls.