Python can’t stop, won’t stop evolving. Python 3.15, due in fall 2026 but available now in its first release candidate, abounds with useful new features worth trying out: lazy imports, a new profiling technology, immutable dictionaries, and its ever-evolving JIT (just-in-time) compiler. Top picks for Python readers on InfoWorld Speed boost your Python programs with new lazy imports Don’t let a fat Python import be a bottleneck for the rest of your program. Lazy imports let you defer executing a module until the time it’s actually used. Best of all, you can use lazy imports with no changes to existing code. Hands-on with the new sampling profiler in Python 3.15 It’s like X-ray goggles for your code! Tachyon, the new profiling system for Python, lets you see what your program is doing without having to instrument it, rewrite it, or even pause it and restart it. Get started with Python’s new frozendict type Python users have been clamoring for an immutable dictionary type, and now they’ve got one — built right into the language. Ever wanted to use a dictionary as a key for another dictionary’s value, or seal a dictionary against modification? Here’s your solution. Get started with Python’s new native JIT We all want faster Python, but nobody wants to rewrite their code. Python’s new built-in JIT is paving the way to automatically faster Python, and you can get started with it right now. More good reads and Python updates elsewhere Run Numba in the browser with WebAssembly Numba speeds up computation in Python by compiling to native code, making it a tricky candidate for porting to an in-the-browser experience. Nevertheless, a new project from QuantStack, called JupyterLite, not only brings Numba into the browser but adds performance boosts too. Mojo is now open source The one-time Python challenger for the scientific computing space has shaped up to be a “Rust lite” language . And now it’s freely available under an Apache 2 license for your tinkering and forking pleasure. Scaling NumPy on free-threaded Python As Python’s free-threaded build gains traction, developers of libraries such as NumPy are working to squeeze maximum oomph from Python’s newfound parallelism. Learn how hidden bottlenecks in NumPy kept it from scaling, and what it took to break that stranglehold. How a common string method in Python turned out to have a security vulnerability When does converting a string into lowercase expose you to danger? When a corner case in how the operation is processed can deliver inconsistent results. Read about how this tiny but troubling problem was detected and fixed.