Releasing Python 3.15.0
This is Hugo van Kemenade's live-tooted account of releasing Python 3.15.0, covering the blockers, rc3 delay, and final steps from October 1 to 9, 2026. It's a behind-the-scenes look at how a CPython release actually gets shipped.
- The final release was pushed back by a surprise rc3, because last-minute lazy-import fixes needed time for proper testing.
- Early blockers included docs.python.org and buildbot outages from an incompatible pyOpenSSL upgrade, and a flaky test that was fixed by raising a timeout.
- The release process has many manual gates, including Windows installer signing that needs button clicks, so no release can be signed without the whole release team being notified.
- Van Kemenade followed a seven-day cooldown for dependencies, sticking with 3.14 for release tooling instead of the new rc3 despite pydantic and aiohttp adding 3.15 support.
Hugo van Kemenade, the release manager, live-toots the path to Python 3.15.0 in a blog post that reads like a running diary of a release cycle. The plan slipped once. The original final date was postponed when a last-minute set of lazy-import fixes landed, prompting a "surprise" third release candidate, rc3, on October 2, 2026. That was followed by a second release-day attempt on October 9. Along the way the release hit a few snags: an expired certificate from an incompatible pyOpenSSL upgrade took docs.python.org and buildbot.python.org offline on October 1 until the US side came online, a flaky test required a restart, and the changelog tool blurb choked on the lack of new NEWS fragments since rc3. That last one needed a fix merged and released to PyPI, a lockfile update, and a backport before the release script would run.
The post is also a useful look at how much ceremony surrounds a CPython release. Windows installers take roughly two hours to build, the Android and iOS artifacts and source tarballs about thirty minutes, and Windows binaries pause for manual approval, with the whole release team emailed each time so no one can quietly ship a signed build. Sigstore signing, SBOM generation, and the python.org uploads come after that. Van Kemenade also deliberately held back from using rc3 to run the release scripts, because he was respecting a seven-day cooldown on dependencies and stuck with 3.14 for tooling. He asks readers whether they use cooldowns too, which is a fair question for anyone pinning fresh releases in production.
For a reader, the practical takeaways are the feature list in the release announcement (lazy imports, a frozendict and sentinel builtin, the Tachyon profiler, UTF-8 as the default encoding, stable ABI for free-threading, TypedDict and TypeForm support, and a faster JIT) and the note that library maintainers should add 3.15 to their test matrices. The post also shows what happens after the release: updating PEP 790, the devguide, and version labels, unlocking the 3.15 branch so about 185 deferred PRs can merge, and handing off to the next release manager for the 3.16 alpha.