PyCharm 2026.2.3

What's new in 2026.2.3

This update fixes several type-checking false positives along with issues in Ruff, Jupyter, Databricks, and interpreter setup:
Code Insight
- Get correct type inference from Numpy-style docstrings, including favoring type annotations over docstring types and handling tuple[...] with three or more arguments - PY-32793 and PY-89314.
- Avoid false positive warnings about missing str or repr implementations for IPv4Address/IPv6Address, string literals, and str subtypes - PY-89986, PY-91292, and PY-89483.
- Avoid a false positive Expected type 'TypedDict | None', got 'dict' instead error when passing a dict where a TypedDict | None is expected - PY-88391.
- Get better handling of # type: ignore comments - PY-90290.
- Avoid duplicated type checker messages for Pydantic Field - PY-90614.
IDE and Tooling
- Avoid a crash loop when using Ruff version 0.16.5 or later - PY-92208.
- Generate a new Python environment with uv even when UV_PROJECT_ENVIRONMENT is set - PY-83666.
- See Jupyter variables again when using a remote Conda interpreter - PY-87590.
- Let the Databricks plugin respect your proxy settings in PyCharm 2026 - PY-91334.
Full release notes.

Full release notes »