Flet Reaches 1.0, Ships One Python App to Six Platforms

What changed in the rewrite

Flet reached version 1.0 on Sept. 15, four years after Feodor Fitsner started the project on the idea that a Python developer should not need a separate frontend stack to ship to iOS, Android, Windows, macOS, Linux and the web. The release replaces the JSON messages Flet previously sent between Python and the underlying Flutter engine with a binary MessagePack protocol, communicating through what the project calls a dart-bridge, and moves Flet's controls onto Python dataclasses.

Fitsner said in the release post that "Flet 1.0 means it's ready for building production apps." The 1.0 branch keeps Flet's original imperative style and adds a declarative option, where a developer describes the interface as a function of the app's state instead of updating controls by hand. Packaged apps can now bundle Python 3.12, 3.13 or 3.14, and Flet's own package index lists more than 100 libraries cleared for mobile builds, including NumPy, pandas, Pillow, SciPy and scikit-learn.

Flet by the numbers at 1.0

Metric	Figure
GitHub stars	16,700+
PyPI downloads	9,000,000+
Contributors	100
Merged pull requests	1,000+
Closed issues	2,800
Discord members	6,000+

Flet's release notes credit the rewrite with up to 6.7 times faster control-diffing than the pre-1.0 release and a redesigned Android packaging path that lets a built app load its APK directly rather than through an intermediate step.

The performance claim nobody outside Flet has measured

RuntimeWire's coverage of the release notes that cross-platform frameworks routinely trade native control, package size or startup speed for developer convenience, and it surfaced a GitHub-reported cold start of roughly 10 to 13 seconds for an Android app built on a pre-1.0 version of Flet. That report predates Sept. 15, so it says nothing about whether the rewrite's packaging and dart-bridge changes fixed it, and Flet has not published its own startup-time figures for the 1.0 release.

A slow first launch is the kind of friction that shows up the moment a tester opens an app on a phone rather than a laptop, which is also where developers have been describing waits of an hour to 16 days just to get a build in front of users at all, on the distribution side of the same six platforms Flet now targets.

What Fitsner isn't saying

Fitsner founded the continuous-integration provider AppVeyor in 2011 before starting Flet in 2022, and RuntimeWire's reporting describes Flet's open-source core as a funnel toward Fitsner's planned paid service, Fletbase. Flet has not published revenue, a paying-customer count, or the number of applications distributed through app stores that were built with it, leaving the 9 million PyPI downloads as a measure of interest rather than of finished, shipping apps.

That gap sits alongside a different kind of open-source math: the one-year donor-funded stretch behind Servo's 1,150 pull-request reviews ran on a budget Phoronix put under $8,000 a month, with no commercial product waiting behind it. RuntimeWire's reporting does not give Fletbase a launch date or a price.