Python 3.15 Is Here: Lazy Imports, frozendict and More

Why this release matters

Python 3.15 was officially released on October 9, 2026. Unlike many version bumps, it brings changes you can see in a few lines of code: explicit lazy imports, a built-in immutable mapping called frozendict, */** unpacking inside comprehensions, and further gains in the experimental JIT compiler.

Lazy imports

Normally a module is loaded the moment its import statement executes. In 3.15 you can mark an import as lazy so the real loading happens only when the name is first used.

import heavy_module
print("Application started")  # heavy_module already loaded

lazy import heavy_module
print("Application started")  # not loaded yet
heavy_module.run()             # loaded here, on first use

Who benefits

• CLI tools where most sub-commands never touch most dependencies.
• Serverless functions and short-lived processes sensitive to cold-start time.
• Large applications and developer tooling with deep import graphs.

frozendict: a built-in immutable dictionary

3.15 adds a dictionary-like type that cannot be modified after creation.

config = frozendict(environment="production", retries=3)

config["retries"] = 5   # TypeError: object does not support item assignment

• Configuration that must never change at runtime.
• Hashable mappings (when keys and values are hashable) — usable as dict keys or in sets.
• Safer sharing of state across modules and threads.

Unpacking inside comprehensions

You can now use * and ** directly in comprehensions, which makes flattening and merging read naturally.

lists = [[1, 2], [3, 4], [5]]
result = [*items for items in lists]
print(result)   # [1, 2, 3, 4, 5]

# Previously
result = [x for items in lists for x in items]
# or: list(itertools.chain.from_iterable(lists))

It's a readability improvement — new syntax, not a performance guarantee.

The JIT keeps improving

The official notes report roughly a 7–8% geometric-mean improvement for the JIT over the standard interpreter on x86-64 Linux, and about 11–12% over the tail-calling interpreter on AArch64 macOS. These are benchmark-suite results, not a promise for every workload.

Should you upgrade?

Try it in development and CI first. Python 3.15 is very new, so verify that your dependencies, framework and deployment images support it before moving production.

Official documentation

• Python 3.15.0 release page (python.org)
• What's New In Python 3.15 (docs.python.org)
• PEP 810 — Explicit lazy imports
• PEP 814 — Add frozendict built-in type
• PEP 798 — Unpacking in comprehensions
• PEP 744 — JIT compilation
• PEP 790 — Python 3.15 release schedule