How do I start a Python REPL with uv?

Run `uv run python` from your project directory. uv creates or reuses the project's virtual environment, installs any required dependencies, and launches an interactive Python session. Outside a project, `uv run python` still works and uses an ephemeral environment.

How do I use a specific Python version in the uv REPL?

Pass `--python` to `uv run`: `uv run --python 3.12 python` starts a Python 3.12 REPL. If that version is not already installed, uv downloads and installs it automatically before launching the session, so no separate `uv python install` step is needed.

How do I run a Python REPL with extra packages using uv?

Use one or more `--with` flags: `uv run --with pandas --with numpy python` creates an isolated environment containing pandas and numpy, then launches Python. The packages stay scoped to that single REPL invocation and never touch the project's `pyproject.toml` or lockfile.

How do I run a Python REPL that ignores the current project's dependencies?

Pair `--no-project` with `--isolated`: `uv run --no-project --isolated python`. `--no-project` alone skips project discovery but still uses a `.venv` sitting in the directory, so the project's packages remain importable. Adding `--isolated` builds an ephemeral environment instead, which is what you want for reproducing issues against the standard library.

How to Run a Python REPL with uv

by Tim Hopper · Markdown

uv verified with uv 0.12

uv does not currently officially support a system-level installation of Python, however, you can try it in preview mode.

To start an interactive Python REPL (Read-Eval-Print Loop) session using uv:

$ uv run python

This command will:

Create or use an existing virtual environment if in a project (i.e. the project has a pyproject.toml file)
Install any required dependencies
Launch an interactive Python session

Using a Specific Python Version

To run the REPL with a particular Python version:

$ uv run --python 3.12 python

If the requested Python version is not installed, uv will download and install it automatically.

Running the REPL with Additional Dependencies

To run Python with specific packages available:

$ uv run --with pandas --with numpy python

This will create an isolated environment with the requested packages installed before launching the REPL.

Running the REPL Isolated from a Project

When working in a project directory but wanting to run a clean REPL without project dependencies:

$ uv run --no-project --isolated python

--no-project on its own skips reading pyproject.toml, but uv still finds the .venv in the directory and its packages stay importable. --isolated is what forces the ephemeral environment.

If you need regular access to a Python REPL, install IPython as a tool with uv tool install ipython. uv python install also writes versioned shims such as python3.12 into ~/.local/bin, covered in How to Add Python to Your System PATH with uv.

Learn more

uv: A Complete Guide covers what uv does, how fast it is, the core workflows, and recent releases.
How to run the IPython shell in your uv project covers the richer IPython REPL for project work.

Last updated on August 17, 2026

How to Run a Jupyter Notebook with uv How to Run Parallel AI Coding Agents on One Python Project

Please submit corrections and feedback...