Python 3.15 released in October 2023 delivers only minor performance improvements over version 3.14 according to informal benchmark results published by a developer using a personal test suite. The tests compare execution times across multiple interpreters including CPython 3.10 through 3.15, PyPy 3.12, Node.js 26.3 and Rust 1.97.

The benchmark evaluates two core algorithms: a recursive Fibonacci calculator and a bubble sort implementation. Each runs in single-threaded and four-threaded modes, with results averaged across three trials. Speed differences are expressed as multipliers relative to Python 3.15, where 1.0x indicates identical performance.

In single-threaded Fibonacci testing, Python 3.15 completed the task in 6.94 seconds compared to 7.17 seconds in 3.14. The JIT-enabled 3.15 version showed a 20% improvement over its standard counterpart. PyPy 3.12 remained the fastest interpreter at 1.25 seconds, outperforming Node.js and Rust in this specific test.

Bubble sort performance followed a similar pattern, with Python 3.15 completing 10,000-element sorts in 1.97 seconds versus 2.06 seconds in 3.14. The JIT version of 3.15 demonstrated a 28% speed gain over its standard counterpart. Free-threading variants showed comparable results to standard interpreters, with no significant regression observed.

Multi-threaded tests revealed more pronounced differences. The free-threading 3.15 interpreter completed Fibonacci calculations in 7.24 seconds versus 32.54 seconds in standard 3.15, achieving a 4.5x speedup. This improvement stems from the removal of the Global Interpreter Lock in free-threading builds, enabling true parallel execution.

Comparisons across Python versions indicate that performance gains have been most significant between 3.10 and 3.11, and between 3.13 and 3.14. Subsequent updates including 3.15 show only marginal improvements, with some tests revealing minor regressions. The JIT feature, introduced in 3.13, continues to show promise but remains experimental.

The benchmark designer cautions against drawing broad conclusions from these results. While Python 3.15 shows measurable gains in specific scenarios, particularly with JIT-enabled workloads, the changes are not substantial enough to warrant immediate upgrades for most production environments. The developer plans to continue using 3.14 for primary projects while adopting 3.15 for personal experimentation, citing new language features like unpacking comprehensions and lazy imports as additional incentives for testing the newer version.

These findings reflect ongoing efforts to optimize Python's execution model while maintaining backward compatibility. The results underscore that performance variations across versions are often context-dependent, with certain workloads benefiting more than others from interpreter improvements.