How the Shannon Plan Targets a 5× Speed Boost for Python

In autumn 2020 CPython core developer Mark Shannon proposed a series of performance improvements known as the “Shannon Plan”.

He launched the Faster CPython project with the goal of making Python five times faster within four years.

Microsoft later joined the effort, and developers including Mark Shannon and Guido van Rossum are working on the “Faster CPython” research.

Recently Shannon and Michael Droettboom outlined the plan for Python 3.13, which includes three parallel work streams:

Tier 2 optimizer – aim to cut interpreter time by at least 50% by getting the Tier 2 interpreter running, generating superblocks, and implementing basic superblock management.

Enable sub‑interpreters from Python code (PEP 554) – builds on the per‑interpreter GIL introduced in Python 3.12, allowing better parallelism without C extensions; a draft PEP 554 is pending approval.

Memory‑management optimizations – reduce allocation overhead and shorten cyclic‑GC pauses by improving data structures.

The official JIT is still a longer‑term goal; the first step is a tracing interpreter, but a full JIT compiler may not appear until Python 3.13, and Shannon remains skeptical about its necessity.

Some developers criticize the internal struggle with another project that seeks to remove CPython’s global interpreter lock (GIL) entirely, noting that the Faster Python team appears to have a stronger political position in that debate.