Python 3.15 released with lazy imports, unpacking in comprehensions and a built-in sampling profiler; Python 3.10 reaches end of support

Python 3.15 has been released as a stable version. It adds explicit lazy imports (PEP 810), unpacking inside comprehensions (PEP 798) and a sampling profiler called Tachyon (profiling.sampling) in the standard library. The tracing profiler cProfile is now named profiling.tracing. Python 3.10 is now officially unsupported. The source is Simon Willison's blog, which points to the official What's New guide. It is not the announcement itself, and the extraction lists no primary links.

Maintainers of libraries and tools can now drop Python 3.10 and rely on 3.11 features such as tomllib, TaskGroup and the SQLite improvements. Willison says he will do this in his own projects. Lazy imports should help programs with heavy dependencies start faster, for example plugin-based command line tools. The built-in sampling profiler means one less third-party install for profiling. This is a routine yearly release, not a structural shift. Most teams will not need to act until their dependencies and deployment platforms support 3.15.

Little of the coverage is hype, but the framing around dropping 3.10 is one person's plan, not an ecosystem-wide move. Many users stay on older Python for years via distribution packages. The only performance figure is a 1.2x gain from the experimental JIT on a single Fibonacci test, which says little about real workloads. Overall speed is roughly the same as 3.14.

1. 01How quickly major libraries (NumPy, pandas, Django, PyTorch) publish 3.15 wheels, and how many drop 3.10 support in their next releases.
2. 02Independent benchmarks of the JIT and of lazy imports on real applications, to test whether the startup and speed gains hold beyond toy tests.
3. 03Early bug reports and breakage tied to the cProfile rename to profiling.tracing and to lazy import behaviour, plus adoption figures for 3.15 in tools like uv and in CI matrices over the next few months.