pandas 3.0 Lands Breaking Changes and Other Python News for February 2026

Last month brought exciting performance news for Python! The Python 3.15 alphas showed JIT compiler speed gains of up to 7–8% on some platforms, while pandas released version 3.0, delivering the most significant performance improvements in years. The Python Software Foundation also received considerable investments in Python security and launched the 2026 Python Developers Survey, and PyTorch 2.10 deprecated TorchScript.

Time to dive into the biggest Python news from the past month!

Python Releases and PEP Highlights

Last month brought two Python 3.15 alpha releases in quick succession, with notable JIT compiler improvements showing promising performance gains. Several PEPs also emerged, including one proposing a cleaner way to write multiline strings.

Python 3.15.0 Alpha 4 and 5: Two Releases in Two Days

January saw an unusual situation in Python’s release history: Python 3.15.0a4 arrived on January 13, but it was accidentally compiled against outdated source code from December 2025. The release team quickly followed up with 3.15.0a5 on January 14 to correct the issue.

Both releases continue the work on Python 3.15’s headline features:

• UTF-8 as the default text encoding for files that don’t specify an encoding, via PEP 686
• A new statistical sampling profiler that’s high-frequency and low-overhead, via PEP 799
• The PyBytesWriter C API for creating bytes objects more efficiently, via PEP 782
• Enhanced error messages with improved clarity and usefulness

The most exciting news for performance enthusiasts is the continued progress on Python’s experimental JIT compiler. Alpha 5 reports a 4–5% performance improvement on x86-64 Linux and a 7–8% speedup on AArch64 macOS compared to the standard interpreter.

If you maintain packages, now is a good time to start running tests against the alphas in a separate environment so you can catch regressions early.

PEP 822 Drafted: Dedented Multiline Strings (d-strings)

A new PEP emerged in January that could make writing multiline strings clearer. PEP 822, authored by Inada Naoki, proposes adding dedented multiline strings (d-strings) to Python.

If you’ve ever written a multiline string inside an indented function or class, you’ve likely run into the awkward choice between breaking your code’s visual structure or using textwrap.dedent() to clean up the extra whitespace. PEP 822 offers a cleaner solution with a new d prefix:

def get_help_message():
# Current approach
return textwrap.dedent("""\
 Usage: app [options]
 Options:
 -h Show this help message
 -v Enable verbose mode
 """)
# Proposed d-string approach
return d"""
 Usage: app [options]
 Options:
 -h Show this help message
 -v Enable verbose mode
 """

The d prefix tells Python to automatically strip the common leading whitespace from each line, using the indentation of the closing quotes as a reference. This differs slightly from textwrap.dedent(), which uses the least-indented line to determine how much to strip.

The proposal targets Python 3.15 and is currently in draft status. If you work with templates, SQL queries, or any code that embeds multiline text, this feature could simplify your workflow. You can follow the discussion on the Python Discourse and provide feedback while the PEP is still being refined.

PSF News: Investments, Fellows, and Survey

The Python Software Foundation (PSF) had a busy month with a major security investment announcement, new Fellows recognition, and the launch of the annual developers survey.

Anthropic Invests in Python Security