The Python implementation written in a Python-derived language and equipped with a just-in-time compiler is PyPy. Its interpreter is written in RPython, a restricted and statically analyzable subset of Python, rather than ordinary application-level Python.
PyPy uses meta-tracing to generate a tracing JIT compiler. Instead of compiling every operation immediately, the system observes frequently executed paths while a program runs and specializes those “hot” paths. This can make pure-Python programs substantially faster than they are under CPython, the reference implementation.
PyPy grew out of research into dynamic-language implementation and was influenced by the earlier Psyco project. It is not simply a faster version of CPython: programs that depend heavily on CPython-specific C extensions may require compatibility layers or may perform differently. The name PyPy reflects its self-referential design—Python-related tools help build a Python implementation.