Command referenceconda install
conda install command (Linux & Windows / Conda)
What it does
Install packages into the active or named Conda environment from configured channels. When people look up conda install, they usually want safe flags, copyable examples, and how this Conda subcommand fits Anaconda or Miniconda workflows on Linux, Windows, and macOS. Syntax, options, and practical examples are below.
Install packages into the active or named Conda environment from configured channels.
Syntax
conda install [-n ENV] PACKAGE [PACKAGE ...] [options]
Common options
| Flag / option | Meaning |
|---|---|
| -n, --name | Target environment name. |
| -c, --channel | Extra channel to search (e.g. conda-forge). |
| --file | Install from a package list / explicit spec file. |
| --revision | Revert the env to a previous revision. |
| -y, --yes | Skip confirmation. |
| pip | Install the pip package into the env when needed for PyPI-only deps. |
Practical examples
Install one package
conda install numpy
Adds NumPy to the active environment.
Pin a version
conda install pandas=2.2
Requests pandas 2.2.x subject to the solver.
conda-forge channel
conda install -c conda-forge jupyterlab
Pulls JupyterLab from conda-forge.
Into a named env
conda install -n ml scikit-learn
Installs without activating ml first.
From a file
conda install --file requirements.txt
Installs each Conda-compatible spec listed in the file.
Multiple packages
conda install matplotlib seaborn
Resolves and installs both packages together.
Linux ↔ Windows equivalent
For pure Python wheels you may still use pip inside an activated Conda env; prefer Conda packages when native libraries are involved.
Common mistakes
- Running `pip install` for everything inside a Conda env without need — prefer Conda packages when available for binary deps.
- Installing into base for every experiment; create project environments instead.
- Adding random channels without pinning priority, which can produce mixed builds.
Related commands
- conda createCreate a new isolated Conda environment with an optional Python version and starter packages.
- conda updateUpdate Conda packages (or Conda itself) to newer compatible versions.
- conda removeUninstall one or more packages from a Conda environment (or remove an entire env with --all).
- conda searchSearch configured channels for available package versions before installing.
- conda listList packages installed in the active or named Conda environment.
- conda activateActivate a named or path-based Conda environment in the current shell session.
- conda env listList all Conda environments and mark the currently active one.
- aptSearch, install, update, and remove packages on Debian-based systems.
FAQ
What does conda install do in Conda?
Install packages into the active or named Conda environment from configured channels.
Does conda install work on Windows and Linux?
Yes. Conda runs on Linux, Windows, and macOS; shell activation differs slightly after `conda init`, but the CLI flags are the same.
What is a practical conda install example?
Try `conda install numpy` — Adds NumPy to the active environment.
Is mamba the same as conda install?
Mamba is a faster drop-in solver for many Conda workflows. Prefer `mamba` only when it is installed; otherwise use the Conda commands on this page.
Browse the full Linux and Windows command reference to search more bash, CMD, PowerShell, Conda, and GitHub CLI commands.