Switching to uv for Python Management

For years, I relied on the official installer from python.org to manage my Python versions. Recently, however, I uninstalled everything to start fresh with uv. Here is how (and why) I made the switch.

Installation

Getting started is simple.

On macOS and Linux:

curl -LsSf https://astral.sh/uv/install.sh | sh

On Windows:

powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

Keeping uv Updated

To ensure you have access to the latest Python releases, you must keep uv itself updated. The list of installable Python versions depends directly on the version of the tool you are running.

uv self update

Installing Python

You can view a list of all available Python builds via the terminal:

uv python list

However, the output can be overwhelming, making it difficult to spot the latest release at a glance. It is often easier to visit python.org to verify the current version number.

Once you have the version number, you can install it. I prefer using the default flag, which adds the installed version to your PATH, allowing you to execute python globally.

uv python install --default --preview-features python-install-default 3.14.3

Note: Replace 3.14.3 with your desired version.

Usage

The biggest shift in workflow is mental: forget about using python or pip directly. From now on, uv handles the heavy lifting.

Initialize a project:

uv init

Install packages (replaces pip install):

uv add numpy

Run a script:

uv run main.py

Why I Made the Switch

I moved away from the official installers for three main reasons:

1. I often need different Python versions for different projects, which the standard installer handles poorly.
2. I used to accidentally install packages to my global workspace. While I knew how to use python -m venv, I would occasionally forget to activate the environment before running pip install.
3. uv streamlines package management. It removes the need to manually activate virtual environments or freeze requirements files. It manages dependencies automatically with fewer commands.

Photo by Juliana Tanchak on Unsplash.