Python 3.15 released, 3.10 support ends

Summary

Python 3.15 introduces explicit lazy imports via PEP 810, derivation opening operations through PEP 798, and built-in sampling profilers named Tachyon for the open-source community. Alongside this release, Python 3.10 officially reaches the end of its lifecycle.

Why it matters

Explicit lazy imports directly eliminate startup costs that can take seconds in plugin-based command-line tools. Willison personally experiences and documents this issue in his own LLM project. Tachyon profilers remove the dependency on external tools like py-spy for low-cost runtime analysis. Additionally, the end of 3.10's lifecycle enables the broader open-source ecosystem to adopt the one-year feature set of 3.11, including tomllib, asyncio TaskGroup, and significant SQLite API additions.

Highlights

• Explicit lazy imports (PEP 810) reduce plugin startup latency by ensuring modules are loaded only upon first use.
• Tachyon sampling profilers are included in the standard library, capable of attaching to any running PID and generating flame graphs and Firefox-profiler output.
• The end of 3.10's lifecycle allows maintainers to adopt features from 3.11, such as tomllib, TaskGroup, datetime.UTC, and new SQLite window function/blob APIs.

Critical take

The primary result of the experimental JIT is a 1.2% speedup in a single Fibonacci benchmark. Willison highlights this as the first 'meaningful' gain after years of iterative work. However, since this feature has been on the roadmap since 3.13, the result appears far from meeting expectations. Beyond this, the article is largely a feature list. There is a near-total lack of critical analysis regarding the costs of these choices, backward compatibility risks, and how these additions shape the Python tooling ecosystem.