Python 3.15 Alpha 6: A Developer Preview Packed with New Features and Performance Gains

Python 3.15 is still under active development, and the latest pre-release, Python 3.15.0a6, is now available for testing. This is the sixth of eight planned alpha releases, designed to give developers an early look at upcoming features and to help refine the release process. Please note that this is a preview release and is not recommended for production environments. Features may still be added, modified, or removed before the beta phase begins on 2026-05-05 and the release candidate phase on 2026-07-28.

New Features and Improvements

Python 3.15 brings several major enhancements and performance upgrades. Here are the highlights so far:

Python 3.15 Alpha 6: A Developer Preview Packed with New Features and Performance Gains

PEP 799: Statistical Sampling Profiler

A new high-frequency, low-overhead profiler based on statistical sampling is being introduced, along with a dedicated profiling package. This will make it easier to diagnose performance bottlenecks without significantly slowing down your code.

PEP 798: Unpacking in Comprehensions

You can now use the * and ** unpacking operators inside comprehensions, making it simpler to combine and transform iterables and mappings within list, dict, and set comprehensions.

PEP 686: UTF-8 as Default Encoding

Python now uses UTF-8 as the default encoding for text files, aligning with modern standards and reducing cross-platform encoding issues. This change applies to open() and other text I/O operations.

PEP 782: PyBytesWriter C API

A new C-level API, PyBytesWriter, allows developers to efficiently create bytes objects in C extensions, providing a simpler and more performant way to build byte sequences.

PEP 728: TypedDict with Typed Extra Items

The TypedDict type hint now supports specifying the type for extra (unlisted) keys, giving you finer control over dictionary shapes without losing type safety.

JIT Compiler Upgrades

The just-in-time (JIT) compiler has received significant upgrades, resulting in a 3–4% geometric mean performance improvement on x86-64 Linux over the standard interpreter, and a 7–8% speedup on AArch64 macOS compared to the tail-calling interpreter. These gains are particularly beneficial for long-running or compute-heavy applications.

Improved Error Messages

Error messages have been refined to be more informative and actionable, helping developers quickly identify and fix issues.

(If you're a core developer and notice a missing feature, let Hugo know — see the resources below.)

Schedule and Resources

The next pre-release, Python 3.15.0a7, is currently scheduled for 2026-03-10. Keep an eye on the official release schedule for updates.

Acknowledgments

Many thanks to all the volunteers who contribute to Python development and release efforts! Your work makes these releases possible. Consider supporting the community by volunteering or through organisational contributions to the Python Software Foundation.

This release was prepared by Hugo van Kemenade, Ned Deily, Steve Dower, and Łukasz Langa.