NEW YORK — What happened: Python 3.15 development releases are emphasizing the stabilization of the 'free-threaded' build, which removes the historic Global Interpreter Lock (GIL) to allow true multi-core CPU utilization in a single Python process.
Why it matters: For decades, Python's GIL has been a bottleneck for CPU-bound tasks, forcing developers to rely on complex multiprocessing setups or external C extensions. Removing the GIL changes how data science, web servers, and concurrent automation are architected in Python.
Deep dive: The removal of the GIL introduces a new set of challenges, chiefly thread safety in legacy C extensions and standard library data structures. Python 3.15 introduces fine-grained locking and biased reference counting to minimize performance regression in single-threaded workloads while unlocking linear scaling across multiple CPU cores for parallel algorithms.
Report check (claims vs what is verified vs still rumor): Benchmarks from the Python enhancement proposal discussions verify near-linear scaling for heavy compute workloads without the GIL. Rumors that all third-party packages like NumPy and Pandas work seamlessly without modification are false; many require explicit updates to ensure thread safety under the new execution model.
Open questions: When will major Linux distributions and cloud providers make the free-threaded build of Python the default installation? How long will it take for the broader package ecosystem to drop GIL-dependent assumptions?