Python 3.15 RC debuts with JIT boosts and lazy imports

Python 3.15 has officially entered the release candidate phase, with the 3.15.0rc1 version debuting on August 4, 2026. The final stable release is expected in October 2026, bringing a suite of performance optimizations and language refinements designed to reduce execution overhead.

At the center of this release is an upgraded Just-In-Time (JIT) compiler, which delivers a geometric mean performance improvement of 8% to 13% over standard CPython. To help developers optimize this increased speed, Python 3.15 introduces Tachyon, a new statistical sampling profiler housed within a dedicated 'profiling' module. The update also implements several highly requested language features, including unpacking in comprehensions (PEP 798) and a built-in 'frozendict' type (PEP 814) to provide a native immutable mapping.

The push for efficiency

These updates follow Python 3.14, which laid the groundwork for the JIT and experimented with an incremental garbage collector. However, the transition has not been without course corrections. Python 3.15 will revert to the older generational garbage collector used in Python 3.13. This decision comes after the incremental collector introduced in version 3.14 was found to increase process memory usage, prompting a return to the more memory-efficient previous model.

Reducing startup overhead

One of the most significant additions for developers is the introduction of lazy imports via PEP 810. This feature allows modules to be processed only at the moment they are actually used rather than at the start of the program. By deferring this processing, Python 3.15 significantly reduces program startup time, addressing a long-standing criticism regarding the language's execution overhead in large-scale applications.

Industry implications

The combination of JIT gains and lazy imports directly targets Python's historical struggle with execution speed. Furthermore, the addition of native immutable structures like frozendict and sentinel types reduces the reliance on third-party libraries and eliminates the need for repetitive boilerplate code when implementing common immutable patterns. This version stands as one of the most feature-packed releases in recent history.

What to watch

As the community moves toward the October final release, the primary focus will be on the real-world stability of the upgraded JIT compiler across different hardware architectures. Developers should monitor the release candidate phase to ensure that the revert to the generational garbage collector successfully resolves the memory spikes observed in version 3.14.