Installation

Flet is a Python library plus a command-line tool - flet create, flet run, flet build, flet test. What you have to install depends on the package manager you use: with uv, nothing at all.

Prerequisites

Python version

Flet requires Python 3.10 or later. Check the version you have with python --version.

Operating System

macOS

Flet supports macOS 12 (Monterey) or later.

Windows

Flet supports 64-bit version of Microsoft Windows 10 and Windows 11.

Linux

Flet supports Debian 10, 11 and 12 and Ubuntu 20.04, 22.04 and 24.04 LTS.

Desktop flavor (audio & video support)

On Linux, the Flet desktop client is available in two flavors: full and light.

The light flavor (default on Linux) does not include audio and video extensions, resulting in a smaller download. The full flavor bundles audio and video support out of the box.

To select the flavor, either set the FLET_DESKTOP_FLAVOR environment variable:

export FLET_DESKTOP_FLAVOR=full

or add the setting to your project's pyproject.toml:

[tool.flet]
desktop_flavor = "full"

If you use the light flavor and need audio or video, you will need to install the required libraries yourself - see the Audio and Video setup guides.

Windows Subsystem for Linux (WSL)

Flet apps can be run on WSL 2 (Windows Subsystem for Linux 2).

However, if you are getting cannot open display error follow this guide for troubleshooting.

Set up your package manager

uv

uv is "an extremely fast Python package and project manager, written in Rust". Install it if you haven't already:

# macOS and Linux
curl -LsSf https://astral.sh/uv/install.sh | sh

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

That is the only thing you need to install - there is no separate step to install Flet, and no virtual environment to create or activate by hand:

• uvx runs the Flet CLI in a temporary, throwaway environment, which is all you need to scaffold a new app.
• Once the app exists, uv run installs its dependencies from the generated pyproject.toml into the project's own virtual environment.

Check that it works:

uvx --with flet-cli flet doctor

The --with flet-cli part is required: the flet package provides the flet command, but the commands themselves live in the separate flet-cli package.

Optional: a global flet command

If you'd rather type flet than uvx --with flet-cli flet, install the CLI as a persistent uv tool:

uv tool install --with flet-cli flet
flet doctor

Upgrade it with uv tool upgrade flet, and remove it with uv tool uninstall flet.

This is handy for commands that act on a project from the outside, such as flet create. To run or build an app, prefer uv run flet ... from inside the project, so that the app uses its own pinned dependencies rather than the global ones.

pip

pip has no equivalent of uvx, so Flet has to be installed into a virtual environment that you create and activate yourself.

Create a directory for your app, and a virtual environment inside it:

mkdir my-app
cd my-app
python -m venv .venv
source .venv/bin/activate

Then install Flet and check that it works:

pip install 'flet[all]'
flet doctor

Every flet command in these docs assumes this virtual environment is activated. You will need to activate it again in each new terminal session.

What gets installed

flet[all] is shorthand for the four packages that make up a complete Flet installation:

Package	What it does
flet	The library you import flet as ft, and the flet command itself.
flet-cli	The implementation of flet create, run, build, test and the other CLI commands.
flet-desktop	The desktop client that flet run opens your app in.
flet-web	The web client and server behind flet run --web and flet serve.

When you don't need all four, install them through individual extras instead: flet[cli], flet[desktop], flet[web], or flet[test] for the integration testing dependencies. A bare flet gives you the library only, which is what a deployed app usually needs.

note

If you run a flet command and flet-cli is missing from the current environment, Flet installs the matching version for you before continuing. The same happens for flet-desktop and flet-web the first time an app runs as a desktop or web app.

Adding Flet to an existing project

If you already have a project and just want Flet in it:

uv

uv add flet
uv add --dev flet-cli flet-desktop flet-web 'flet[test]'

The first command makes flet a runtime dependency of your app. The second puts the tooling in the dev dependency group, so it is available to uv run flet ... without shipping with your app. This is the same layout that flet create generates.

pip

pip install 'flet[all]'

Unlike uv add, pip does not record anything: add flet to your requirements.txt or pyproject.toml yourself.

Installing a pre-release version

Flet publishes pre-release ("dev") builds to PyPI ahead of every stable release, so you can try out upcoming changes early. They are listed on the release history page. Neither uv nor pip will install one unless you ask for it.

uv

Add --prerelease allow to any uv command:

# scaffold an app with the latest pre-release
uvx --prerelease allow --with flet-cli flet create my-app

# or pin an exact version
uvx --prerelease allow --from 'flet[cli]==1.1.0.dev5' flet create my-app

# a global CLI, or an existing project
uv tool install --prerelease allow --with flet-cli flet
uv add flet --prerelease allow

An app scaffolded by a pre-release CLI depends on a pre-release of Flet, so the flag is needed to run it, too:

cd my-app
uv run --prerelease allow flet run

To stop repeating the flag, set it once in the app's pyproject.toml:

[tool.uv]
prerelease = "allow"

pip

Pass --pre, or pin an exact version:

pip install --pre 'flet[all]'

# or
pip install 'flet[all]==1.1.0.dev5'

Upgrading Flet

uv

# your app's dependencies
uv add flet --upgrade

# the global CLI, if you installed one
uv tool upgrade flet

uvx needs no upgrading - it picks up new releases on its own. Add --refresh if it looks like it is using a stale cached version.

pip

pip install 'flet[all]' --upgrade

Next steps

Now you are ready to create your first Flet app.