Python at C speed.
Without rewrites.
pyscriptc compiles pure Python workloads into standalone single executables. No runtime virtual environments. No Docker layers.
Throughput
Higher is betterCompute kernels geometric mean
Peak Memory (RSS)
Lower is betterPeak RSS, kernel group
Standard Library
Higher is betterStandard library workloads geometric mean
Full Python semantics.
Pure machine execution.
Full Python Semantics
CPython 3.14.9 semantics for the supported subset
Write idiomatic, expressive Python for the supported subset. pyscriptc compiles pattern matching and generator expressions directly into native machine code, and the standard library is part of the target.
Predictable, Not Probabilistic
Deterministic execution without JIT warmup cliffs or GC spikes
Compile-Time Diagnostics
Catch type mismatches and fence violations before shipping
def classify(tag: str) -> str:
match tag:
case "stream":
return "streaming"
case "ping" | "pong":
return "keepalive"
return "unknown"Two surfaces today. One unified compiler.
Target the execution surface your infrastructure requires without rewriting a single line of Python. Today that means zero-dependency native binaries, shared libraries with a C surface, and an opt-in island fallback.
Zero dependencies. Machine code. One self-contained file.
Your Python source is compiled down to optimized native assembly. No bundled interpreter, no container bloat, no virtualenv. A single self-contained binary (about half a megabyte) for programs inside the supported subset.
Standalone Single Binary
Ship a self-contained executable with zero runtime dependencies. Drop it into a distroless container, a bare metal server, or an embedded device.
Measured Kernel Speedup
22.8x geometric-mean speedup on compute kernels and 5.3x lower peak RSS than CPython 3.14.9 on Linux x86_64.
Target Every Real Architecture
One command compiles for Linux x86_64 and aarch64, plus aarch64 and x86_64 macOS.
Built for production systems.
Single-command compilation across architectures with no per-target sysroot or SDK setup, pure package support, and documented, numbered divergences.
Deploy Anywhere
Single-command cross-compilation for any cloud or desktop platform. No per-target sysroot or SDK setup.
PyPI Package Support
Compile pure Python packages directly into machine code. Eliminate virtualenv setup, wheel build-matrix hurdles, and runtime dependency bloat in production containers. Top-100 PyPI packages are tracked nightly against the pinned suite; the live breakdown lives on the compatibility site.
CLI-First Compiler
Control the compiler from the terminal or automate inside CI scripts. Full access to IR stages, LLVM output, and diagnostic codes.
Shared Libraries & Headers
--emit binding compiles a typed module into a shared library (.so, .dylib, .dll) plus a ctypes loader, so CPython calls your typed functions without Python.h or an extension build.
CPython 3.14 Parity
CPython 3.14.9 semantics for the supported subset. Write idiomatic standard Python: pattern matching, standard exception handling and generator expressions compile into native machine instructions.
Compile-Time Diagnostics
Deep static type verification and fence boundary checks. Identifies type instability, unboxed scalar leaks, and dynamic fallback barriers before your binary reaches production servers.
Ready to ship Python as native code?
Start compiling standard typed Python into standalone executables in under two minutes.