Python uv Advanced: Python Versions, pip Interface & Tool Installation

When requirements move from "as long as it runs" to "multi-version testing, backward-compatible pipelines, and global tool management," uv can handle it. This article takes you from solo dungeon runs to raiding as a team.

Python Version Management

uv has built-in Python download and management — no need to install pyenv separately.

# Install multiple versions
uv python install 3.10 3.11 3.12
 
# Pin a project to a specific version (creates .python-version)
uv python pin 3.11
 
# Create a venv with a specific version
uv venv --python 3.12
 
# Run with a specific version
uv run --python 3.10 pytest

The first time you use a version, it downloads automatically — subsequent uses go straight from cache.

pip-Compatible Interface

Existing CI, Docker, and scripts still using pip? uv provides a drop-in replacement:

# Equivalent to pip install -r requirements.txt
uv pip install -r requirements.txt
 
# Compile requirements (cross-platform, reproducible)
uv pip compile requirements.in --universal -o requirements.txt
 
# Sync environment (strictly from lockfile)
uv pip sync requirements.txt
 
# Create venv
uv venv

Switch to uv pip and the parameters are almost identical, but the speed jumps by an order of magnitude — like upgrading from a bicycle to a motorcycle: same road, completely different experience. Want to upgrade your legacy project without major changes? Start with uv pip, painless lift-off.

Global Tool Installation (uv tool)

Don't want to run uvx every time for tools you use frequently? Install them globally:

uv tool install ruff
uv tool install httpie
 
# Call directly afterward
ruff check .
http https://httpbin.org/get

Upgrade individual tools or all of them at once:

uv tool upgrade ruff
uv tool upgrade --all

uv tool install creates isolated environments that won't conflict with project dependencies. For project-pinned versions, still use uv add --dev.

Script Lockfile and Shebang

Want a single script to be reproducible too? You can lock its dependencies:

uv lock --script example.py

This creates example.py.lock. When you run uv run example.py, it uses that lockfile.

Or make it executable:

#!/usr/bin/env -S uv run --script
# /// script
# dependencies = ["httpx"]
# ///
 
import httpx
print(httpx.get("https://example.com"))

chmod +x fetch
./fetch

Next Steps

With advanced capabilities mastered, use a set of guidelines to reduce risk: 👉 Best Practices