News · Tools

Python 3.15 Is Out: UTF-8 by Default, Lazy Imports and a Faster Experimental JIT

Python 3.15 shipped October 9, 2026 with UTF-8 as the default encoding and explicit lazy imports, while Python 3.10 reached end of life October 1.

A laptop screen in a dark room shows a Python script for parsing JSON data in a code editor
On this page

The Python core developers released Python 3.15 on October 9, 2026, its new feature release. Two follow-up bug-fix releases came within weeks: 3.15.1 on October 21 and 3.15.2 on October 26. Eight days before 3.15 arrived, on October 1, Python 3.10 reached the end of its five years of support.

What Changed in Python 3.15

The What’s New document lists the biggest changes as lazy imports, two new built-in types (frozendict and sentinel), UTF-8 as the default encoding, unpacking in comprehensions, and a stable binary interface for the free-threaded builds.

Two new built-in types arrive. The documentation describes frozendict as “a new immutable type,” a dictionary that cannot be changed after it is created; trying to change one raises “TypeError: 'frozendict' object does not support item assignment.” Sentinel is a way of “creating unique sentinel values with a concise representation.” The release page also lists improved error messages, and the documentation says the interpreter “now provides more helpful suggestions in AttributeError exceptions when accessing an attribute on an object that does not exist.”

UTF-8 by Default, and Why Windows Users Should Care

The change most likely to touch everyday analysis work is the encoding default. Under PEP 686, “Python consistently uses UTF-8 for default encoding of files, stdio, and pipes.” A text or CSV file opened without naming an encoding is now read as UTF-8.

The proposal says plainly that “this change mostly affects Windows users,” because most other systems already used UTF-8. It also warns that the switch “may cause UnicodeError, mojibake, or even silent data corruption” when a file was saved in an older encoding, mojibake being the garbled characters that appear when text is decoded the wrong way. The fix it suggests is to say what you mean: pass encoding="utf-8" or encoding="locale" when opening a file. Anyone who needs the old behavior back can turn UTF-8 mode off with the PYTHONUTF8=0 environment variable or the -X utf8=0 option.

The change applies to files read through the built-in open() function and the modules that rely on it, such as csv. pandas’ read_csv already assumed UTF-8 by default, so code that loads data through pandas should behave as before.

Lazy Imports in Practice

PEP 810 adds a lazy soft keyword, written as lazy import json or lazy from json import dumps. In the proposal’s words, “Lazy imports defer the loading and execution of a module until the first time the imported name is used.” The release page describes the feature as a route to “faster startup times.”

There are limits. The keyword is only allowed at the top level of a module, not inside a function, and neither star imports nor future imports can be lazy. Because the module loads later, an error in it now surfaces later too: the documentation says Python “raises the exception at the point of first use rather than at import time.” Our reading: for a notebook or a short analysis script the gain is small, and it matters more for command-line tools that import many heavy libraries but use only a few on each run.

A Faster JIT and a New Profiler

The release page says the experimental just-in-time compiler improved by a 7 to 8% geometric mean over the standard interpreter on x86-64 Linux. It remains experimental. Python 3.15 also adds Tachyon, which the page calls a “high frequency statistical sampling profiler,” inside a new profiling package, and the official macOS installers now include free-threading support by default.

The release page also carries a warning for Mac users: IDLE and tkinter can hang on macOS 27.0, with a link to the open issue. Students who rely on IDLE, the editor bundled with Python, should check that note before upgrading a Mac.

Do pandas and scikit-learn Work on 3.15?

The libraries most analytics courses use already publish packages for the new version. On the Python Package Index, pandas 3.1.1, released November 4, 2026, and NumPy 2.5.5, released November 15, both list Python 3.15 as supported and ship packages built for it. The pandas release notes record that version 3.0.6, from September 17, “is now compatible with Python 3.15.”

scikit-learn 1.9.1 and matplotlib 3.11.2, both from September 2026, still list Python 3.11 through 3.14 in their metadata, but each publishes installable packages built for 3.15. In practice that means the four libraries install on the new version today, though scikit-learn and matplotlib have not yet updated their stated support.

How Long Each Version Is Supported

The 3.15 release schedule, PEP 790, says the version will get bug-fix releases about every second month for two years, with 3.15.3 expected December 8, 2026, then security fixes only until about October 2031. The developer guide’s version table shows Python 3.11 through 3.14 still supported, with 3.11 moving to end of life in October 2027, and lists Python 3.10 as end of life since October 1, 2026, so it no longer receives security fixes.

What This Means for Students

Python is the language programs name most often. Of the 83 programs in our national pool whose curriculum pages name their software, 68 name Python and 58 name R. Employers ask for it too: O*NET’s tally of 2025 job postings shows Python in 66% of data scientist postings, ahead of SQL at 51%, and in 20% of business intelligence analyst postings.

Programs rarely name a version on their curriculum pages, so the practical step is to check your own. If your laptop still runs Python 3.10, move to a supported release before a program starts, because 3.10 now gets no security fixes. Whether to jump straight to 3.15 is a separate call. Install the version your course or employer names; if none is named, the library support above means 3.15 is a workable choice for pandas, NumPy, scikit-learn and matplotlib work. If you read files with open() on Windows, name the encoding, as PEP 686 suggests, so a file saved in another encoding does not come in garbled.

If you are still deciding whether the coding load suits you, our guide to how hard data analytics is covers the math and programming programs expect, and the data scientist career guide shows which tools the job uses most. Programs that lean hardest on Python are in the data science ranking, and business intelligence work, where Python shares the stage with SQL and dashboards, is covered in the BI analyst career guide.

Reported and written by Kevin To. Every figure in this story comes from a source listed below; corrections go to hello@savomediagroup.com.

Frequently Asked Questions

When was Python 3.15 released?

October 9, 2026. Bug-fix releases 3.15.1 and 3.15.2 followed on October 21 and October 26, and 3.15.3 is expected December 8.

Is Python 3.10 still supported?

No. The Python developer guide lists Python 3.10 as end of life since October 1, 2026, so it no longer receives security fixes. Python 3.11 through 3.15 are supported.

Does pandas work with Python 3.15?

Yes. pandas release notes mark version 3.0.6, released September 17, 2026, as compatible, and its current release lists Python 3.15 support. NumPy, scikit-learn and matplotlib also publish packages built for 3.15.

Why do my text files look garbled in Python 3.15?

Python 3.15 reads files as UTF-8 unless told otherwise. A file saved in an older Windows encoding can then decode wrongly. Pass the right encoding when opening it, or set PYTHONUTF8=0 to restore the old default.

Should I upgrade to Python 3.15 for a data analytics course?

Use the version your course names. If none is named, 3.15 works with the main data libraries, but move off Python 3.10 either way, because it no longer gets security fixes.

Sources

  1. Python Software Foundation, Python 3.15.0 release page (opens in a new tab)The release date, the feature list, the JIT figure, the profiler and the macOS warning. · Accessed Nov 18, 2026
  2. Python documentation, What’s New in Python 3.15 (opens in a new tab)The biggest changes, the new built-in types, error messages and lazy import errors. · Accessed Nov 18, 2026
  3. Python Enhancement Proposal 686, Make UTF-8 mode default (opens in a new tab)What the encoding change does, who it affects, the risks and how to opt out. · Accessed Nov 18, 2026
  4. Python Enhancement Proposal 810, Explicit lazy imports (opens in a new tab)The lazy import syntax, behavior and restrictions. · Accessed Nov 18, 2026
  5. Python Enhancement Proposal 790, Python 3.15 Release Schedule (opens in a new tab)The 3.15.1, 3.15.2 and 3.15.3 dates, the bug-fix cadence and the security end date. · Accessed Nov 18, 2026
  6. Python Developer’s Guide, Status of Python versions (opens in a new tab)Which versions are supported and the Python 3.10 end of life date. · Accessed Nov 18, 2026
  7. Python Package Index, pandas, NumPy, scikit-learn and matplotlib project pages (opens in a new tab)Latest versions, release dates, supported Python versions and 3.15 packages. · Accessed Nov 18, 2026
  8. O*NET OnLine, Hot Technologies for Data Scientists and Business Intelligence Analysts (opens in a new tab)The share of job postings naming Python and SQL. · Accessed Nov 18, 2026