Python 3.15 has reached Release Candidate 1, and the final release is expected on October 1, 2026.
I went through the changes and tried to summarize the ones that seem most relevant for everyday Python development without going through all the PEP numbers.
- Faster startup with lazy imports
Python 3.15 introduces lazy imports, which means some modules can be loaded only when they are actually needed instead of being loaded immediately.
This could help applications and CLI tools that spend a noticeable amount of time importing modules before doing any actual work.
- UTF-8 becomes the default
UTF-8 is becoming the default encoding, which should reduce encoding-related problems, especially when code is running on different operating systems.
This should make situations where something works on one computer but fails because of a different system encoding less common.
- Built-in immutable dictionaries
Python 3.15 adds a built-in frozendict-style immutable mapping.
Similar to how a tuple provides an immutable alternative to a list, this gives Python developers a standard way to work with dictionaries that cannot be modified.
- JIT improvements
The JIT compiler continues to improve in Python 3.15.
Early benchmarks show performance improvements in some workloads, including roughly 8–9% on Linux and larger improvements in some Apple Silicon tests.
These numbers will obviously depend on the workload, so I would wait for more benchmarks before making broader conclusions.
- New profiler: Tachyon
Python 3.15 also introduces Tachyon, a new sampling profiler designed for very low overhead.
It can sample running programs at very high frequencies and can also be attached to an already-running process.
This could be useful for finding performance problems without having to restart an application with a profiler attached from the beginning.
- Some terminal improvements
The interactive Python experience is getting a few smaller improvements, including colored error messages and prompts.
The sqlite3 command-line interface also gets SQL keyword completion.
- Free-threaded Python is still opt-in
Python 3.15 does not make the no-GIL/free-threaded build the default.
It is still something you have to explicitly use.
However, free-threaded Python is becoming more mature, including improvements to ABI support that should make it easier for C extension developers to support it.
Overall, Python 3.15 doesn't look like a release that completely changes how Python is written. Most of the changes are focused on performance, tooling, and improving some long-standing parts of the language.
Since this is already RC1, the major feature set should be mostly locked and the remaining work should mainly be bug fixes and final polishing.
Has anyone here been testing the Python 3.15 beta or RC?
I'm especially interested in whether lazy imports have caused compatibility problems with existing projects.
Top comments (2)
This is a great summary of what looks like a very pragmatic Python release. 🚀
What stands out to me is that Python 3.15 seems less about introducing flashy syntax and more about improving the developer experience and runtime itself.
Lazy imports are probably the feature I’m most curious about. The potential startup-time improvements for CLIs, serverless workloads, and large applications are interesting, but I think the compatibility implications will be just as important. Import-time side effects and assumptions about module initialization order could make this something teams need to test carefully before adopting broadly.
I’m also glad to see continued investment in free-threaded Python and the JIT. Those changes could have a much bigger long-term impact on Python performance than any individual language feature.
And having a low-overhead profiler like Tachyon is particularly interesting for production systems, where traditional profiling can sometimes be difficult to introduce safely.
Overall, it feels like Python is continuing to optimize the fundamentals rather than changing the language just for the sake of change.
The real question for me is: which of these improvements will actually make the biggest difference in production applications once Python 3.15 becomes mainstream?
hello Could you tell me about your experience about Python.
how many years of experience have you got?