# Review request for Issue #92810 **URL:** \ **Category:** Core Development\ **Tags:** review-request\ **Created:** [September 9, 2026, 8:39am UTC](https://discuss.python.org/t/review-request-for-issue-92810/108965 "2026-09-09T08:39:49Z")\ **Posts on this page:** 1\ **Page:** 1
**Author:** ![dolfinus](https://sea2.discourse-cdn.com/flex002/user_avatar/discuss.python.org/dolfinus/32/27354_2.png) [@dolfinus](https://discuss.python.org/u/dolfinus)\ **Post date:** [September 9, 2026, 8:39am UTC](https://discuss.python.org/t/review-request-for-issue-92810/108965/1 "2026-09-09T08:39:49Z")
There is an issue [#92810](https://github.com/python/cpython/issues/92810) - `isinstance(ABC_subclass, ...)`/`issubclass(ABC_subclass, ...)` can consume gigabytes of RAM if few thousand unlrelated classes are created and compared with each other. There are 4 related PRs: - additional tests for `isinstance`/`issubclass` cached and non-cached runs, no code changes [https://github.com/python/cpython/pull/144941](https://github.com/python/cpython/pull/144941) - Implementation 1 [https://github.com/python/cpython/pull/131914](https://github.com/python/cpython/pull/131914) - Implementation 2 [https://github.com/python/cpython/pull/141171](https://github.com/python/cpython/pull/141171) - Implementation 3 [https://github.com/python/cpython/pull/150540](https://github.com/python/cpython/pull/150540) The main [question](https://github.com/python/cpython/issues/92810#issuecomment-3621020168) here “should `ABC. __subclasses__ `override influence the behavior of `isinstance`/`issubclass`” - overriding ` __subclasses__ ` for a regular classes (with no metaclass) have no influence in `isinstance`/`issubclass` behavior. If answer is “no”, then Implementation 3 is recommended. I haven’t found [any examples](https://github.com/python/cpython/issues/92810#issuecomment-2765442065) on Github where ` __subclassess__ ` are overriden, other than Mocks and custom metaclasses. A review would be appreciated.