Python's Global Interpreter Lock (GIL) Becomes Optional: Details and Roadmap

Python’s Global Interpreter Lock (GIL) has long limited true multithreaded concurrency, allowing only one thread to execute bytecode on a CPU at a time. Recent discussions indicate that the GIL will become optional, a change hailed as a major win for the AI ecosystem.

PyTorch core maintainer Dmytro Dzhulgakov noted that removing the GIL will enable Python code to run freely across multiple threads, and Meta engineer Sam Gross spent over four years developing the no‑GIL implementation.

Prominent AI researcher Yann LeCun praised the move, emphasizing that Python code will soon be able to execute multithreaded workloads without the GIL’s constraints.

CPython core developer Thomas Wouters outlined the proposal’s details and future outlook, emphasizing three development stages:

Short‑term: an experimental no‑GIL build expected in Python 3.13 (or possibly 3.14), not set as the default.

Mid‑term: after sufficient community support, the no‑GIL build will become an optional default in a future release, with a timeline of one to two years.

Long‑term: the no‑GIL configuration aims to become the default interpreter, fully removing the GIL while preserving backward compatibility, a process projected to take about five years.

The proposal stresses careful handling of backward compatibility, ABI considerations, and the need for community feedback before making the no‑GIL mode the default. The development team will continuously evaluate progress and adjust timelines as needed.