Why uv Is Revolutionizing Python Package Management and Environments

uv is a high‑performance, all‑in‑one tool for the Python ecosystem, written in Rust. It handles package management, environment creation, project initialization, tool execution, and Python version management, offering 10–100× speed improvements over pip and strong compatibility with tools such as pip, virtualenv, pipx, and poetry.

Key Features

Replaces pip, pip‑tools, pipx, poetry, pyenv, twine, virtualenv and more with a single tool.

Runs 10–100× faster than pip.

Provides full project management with lockfile support.

Executes scripts directly and supports inline dependency metadata.

Installs and manages multiple Python versions.

Runs or installs command‑line tools published as Python packages.

Maintains a pip‑compatible interface while delivering superior performance.

Supports Cargo‑style workspaces for large projects.

Efficient disk usage with global cache deduplication.

Installs via a single curl or PowerShell command without pre‑installing Rust or Python.

Cross‑platform: macOS, Linux, Windows.

Installation

For Linux and macOS: curl -LsSf https://astral.sh/uv/install.sh | sh For Windows PowerShell:

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

Project Management with uv

uv automatically creates a virtual environment based on the pyproject.toml in the current (or parent) directory. A minimal pyproject.toml might look like:

[project]
name = "my_project"
version = "1.0.0"
requires-python = ">=3.9,<3.13"
dependencies = [
  "astropy>=5.0.0",
  "pandas>=1.0.0,<2.0",
]

Running uv sync in a directory containing this file will automatically install a compatible Python version, create a .venv with all dependencies, and generate a uv.lock file that pins exact package versions.

Running Commands

Instead of activating a virtual environment manually, use: uv run myscript.py Similarly, launch Jupyter Lab with:

uv run jupyter lab

Adding Dependencies

uv can modify pyproject.toml automatically. To add NumPy version 2 or higher:

uv add numpy>=2.0

Pinning a Python Version

Lock the project to a specific Python interpreter: uv python pin 3.12.9 This ensures the same Python version is used across all machines.

uvx – One‑off Tool Execution

Run a tool without pre‑installing it, e.g., Ruff for linting: uvx ruff Or start an IPython session with pandas and pyarrow pre‑installed: uvx --with pandas,pyarrow ipython uvx downloads the requested tool, caches it, and executes it in a temporary environment.

Overall, uv simplifies and speeds up Python development, providing consistent environments and a unified workflow for both local development and CI pipelines.