Abstract:
With the release of Python 3.15, the Python development environment for the Windows on Arm platform has further matured: when using Python Installation Manager 26.4 to install 3.15 on an Arm device, the native ARM64 interpreter will be selected by default instead of the x64 version. Python 3.15 was officially released on October 9; Microsoft said that Windows on Arm has now become a Tier 2 platform for CPython, which means that the architecture incorporates continuous construction and testing. Related faults need to be repaired in a timely manner, and platform test failures will affect the release of the official version of CPython.
Microsoft statistics show that as of September 2026, among the 1,000 most downloaded projects on PyPI, about 94% of the latest versions have released software packages available for Windows on Arm. However, this ratio does not mean that all projects provide native ARM64 code: the statistics also include native ARM64 wheels, as well as universal Python wheels that do not rely on specific processor architectures and can be used across platforms. Even if the top-level software package is a universal package, the underlying components it depends on may still contain unported native extensions, so it cannot be concluded that every Python application can run completely on Arm computers based on this number alone.
The available tools and libraries listed by Microsoft cover NumPy, SciPy, pandas, scikit-learn, Django, FastAPI, as well as Jupyter, VS Code, Ruff and pytest, etc. The Anaconda main repository has also provided more than 2,500 Windows on Arm software packages, covering Python 3.12 to 3.14, including scientific computing, machine learning, web development and database-related components. For x64 dependencies that have not yet been natively adapted, Windows emulation can still be used as a transitional solution.
Ecological improvements also benefit from the native build environment for developers. GitHub Actions now provides a Windows on Arm managed runner, which allows open source maintainers to compile software packages and run tests directly on Arm machines without having to maintain physical devices or rely on simulation environments. Microsoft said that it will continue to work with maintainers to complete native wheel support for projects such as PyArrow, gRPC, OpenCV and some database clients.
Microsoft views this development as a step towards a more complete development platform for Windows on Arm from “being able to run Python”, but neither Tier 2 support level nor popular package coverage constitutes a guarantee of application compatibility. Developers still need to check the complete dependency tree when migrating projects, especially packages with C, C++ or Rust native extensions, and actually build and test them on the target Arm device.
Comments