A Python with no interpreter and no bytecode, and one week old
The pitch is audacious: make Python what bun and v8 are to JavaScript, a runtime fast enough that the interpreter simply disappears.
Each module is parsed by the ruff parser, lowered to one shared IR, and compiled to native code by Cranelift.
A reviewer of AI-written code saw 'clear signs of vibing' and bet full CPython compatibility won't land on any reasonable timeline.
No interpreter, no bytecode: Python straight to machine code
PON bills itself as a JIT and ahead-of-time native compiler and runtime for Python 3.14, written in Rust, with no interpreter and no bytecode anywhere. The toolchain is unusual: it borrows ruff's parser for the front end, lowers everything to a single intermediate representation, and hands that to Cranelift for code generation, with a Green Tea garbage collector and a NULL-sentinel error model underneath.
The compatibility numbers are early and honestly labeled. Today 209 modules from a differential corpus run byte-identical to CPython 3.14 under the JIT, and 172 of those also match under ahead-of-time compilation. Full CPython test-suite parity is not done. The headline speedups, a geomean of at least 5x and at least 20x on numerics, are roadmap targets with no measured benchmarks attached yet.
'Clear signs of vibing' on a very hard problem
An engineer who reviews LLM-generated code read the repo and saw the pattern immediately:
one week old project, clear signs of vibing. I will be shocked if the remaining work listed (cpython test suite) proceeds in any reasonable timeline. This is a pretty hard problem to just solve in a week.
An embedded software engineer named a concrete wall:
pickle files are usually the limiter here. I would be surprised if it can handle pickle files since it relies so much on runtime LUTs of the objects and arbitrary object definitions. This usually doesn't work in other use cases such as swig or cython either
A critic who worked through the failure modes doubted the whole framing:
It's not Python by any means, it's a subset with its own runtime, its own quirks and nuances… It will be impossible to maintain parity with CPython without AI assistance… It will die the same way as dozens of similar (even non-AI projects) died before
The historical comparison did the rest. As one commenter put it:
Surely this will succeed where $4B Modular failed!
Modular poured billions into making Python fast and still hasn't unseated CPython; a one-week solo project inherits that same graveyard.
Get breakdowns like this every day
The top Hacker News stories, unpacked — by email, Telegram, or RSS.
Amazon stops taking new Mechanical Turk customers as the crowd-work era ends
Amazon will stop accepting new Mechanical Turk customers on July 30, 2026. Current ones can stay, but AWS "does not plan to introduce new features" for the platform that helped invent modern crowd work.