exaloop/codon

Health v2 · maintenance only

Adoption not part of the score

16834 stars · 603 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Codon is a high-performance Python compiler that statically compiles Python-like code to native machine code via LLVM, typically achieving 10-100x speedups over CPython. It supports native multithreading without the GIL, GPU programming, a built-in NumPy implementation, and interoperability with Python, C/C++, and JIT compilation of functions in existing Python codebases.

Use cases

• speed up slow python code without rewriting in c
• compile python to native machine code
• write multithreaded python code without the gil
• accelerate numpy code
• create python extension modules without c or cython
• jit compile performance-critical functions in a python codebase
• run python on embedded or edge devices

When to choose

• your python code is compute-bound and you want C/C++-level performance
• you need native multithreading or GPU execution from Python-like code
• you want to build fast Python extensions without writing C/C++/Cython
• you can accept a mostly-Python language that is not a drop-in CPython replacement

When to avoid

• you need a fully drop-in CPython replacement for a large dynamic codebase
• your code relies heavily on Python's dynamic features or unported modules
• you need native Windows support (use WSL)
• you depend on arbitrary third-party Python libraries at runtime without interop overhead

Facets

cli-tool · maturity active

compiler concurrency gpu-computing cli plugin-system programming-languages compilers performance developer-tools python cpp cli python-compiler ahead-of-time-compilation numpy jit no-gil python-extensions llvm compiler linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="exaloop/codon")