Skip to content
Latchkey

mamba install: Usage, Options & Common CI Errors

mamba install is a much faster, conda-compatible package installer.

mamba is a reimplementation of conda's solver in C++ and is dramatically faster at resolving environments. In CI it is the go-to when conda's solve time dominates the build.

What it does

mamba install accepts the same arguments as conda install but resolves dependencies far faster using the libsolv-based solver. micromamba is an even leaner, standalone binary (no base conda install) ideal for slim CI images. Channels and environments work identically to conda.

Common usage

Terminal
mamba install -y -c conda-forge numpy pandas
mamba install -n myenv -y scipy
micromamba create -y -n ci -c conda-forge python=3.12
micromamba install -y -n ci requests

Common errors in CI

Same channel pitfalls as conda: "nothing provides X" or "PackagesNotFoundError" means the package needs -c conda-forge. With micromamba you must activate or pass -n the target env, or you may accidentally write to base. "mamba: command not found" means it is not installed - use the Miniforge/Mambaforge installer or the micromamba binary. Mixing mamba and conda in the same env can occasionally desync metadata; pick one tool per environment in CI.

Options

FlagWhat it does
-n <env>Target a named environment
-c <channel>Add a channel (conda-forge)
-yAssume yes (required in CI)
micromambaStandalone, no-base lightweight variant

Related guides

Run this faster and cheaper on Latchkey managed runners. Start free →