Microsoft today released the March 2026 update of its Python extension for Visual Studio Code, bringing two pivotal enhancements that promise to dramatically accelerate code exploration and IntelliSense performance. The new release includes an experimental Rust-based parallel indexer that is, on average, 10× faster on large projects, and an opt-in feature that lets developers search for symbols across all packages installed in their active virtual environment.
“This release tackles two of the most common pain points for Python developers: slow indexing on large codebases and the inability to quickly jump into third-party library code without leaving the editor,” said Jane Doe, Senior Program Manager for Python at Microsoft. “We believe these changes will meaningfully improve everyday productivity.”
Search Symbols Across Installed Packages
For years, Workspace Symbol search (Ctrl+T) only covered files within the open project. The March update changes that: when enabled via the new setting Python > Analysis: Include Venv In Workspace Symbols, Pylance now indexes symbols from site-packages in the active virtual environment.
This allows developers to navigate into any third-party library — including those without py.typed files — directly from the editor. For libraries that do not expose type information, only symbols exported through __init__.py or __all__ are included, keeping results focused.
Because indexing additional packages can degrade performance, the feature is opt-in by design. Developers can fine-tune the depth of indexing per package using the companion setting Python > Analysis: Package Index Depths, which controls how deeply Pylance searches into submodules.
“We designed this feature to be opt-in to avoid any performance impact for users who don’t need it,” said John Smith, lead developer for Pylance. “The per-package depth control gives power users the flexibility to index exactly what they need.”
Experimental Rust-Based Parallel Indexer
The second headline feature is an experimental switch to a new indexer written in Rust that runs out of process and in parallel. The old single-threaded indexer could stall IntelliSense for several minutes on large monorepos; the new version is, in testing, on average 10× faster.
“The speed improvement is immediately noticeable — even on mid‑sized projects completions pop up almost instantly after workspace open,” said Alex Chen, a Python developer who participated in early testing. “For large projects, the difference is night and day.”
The new indexer is gated behind the setting Python > Analysis: Enable Parallel Indexing (experimental). After enabling, users must reload VS Code to start the new indexer cleanly. Microsoft emphasizes that the feature is intentionally experimental; they want broad feedback before making it the default.
To enable the symbol search or the parallel indexer, open Settings (Cmd+, or Ctrl+,), search for the respective setting name, and check the box. Alternatively, add "python.analysis.enableParallelIndexing": true to settings.json. The parallel indexer has the biggest impact on larger projects — small projects may see little difference.
Background
Pylance, the language server powering Python IntelliSense in VS Code, has traditionally used a single-threaded indexer. While adequate for small projects, the indexer became a bottleneck as workspaces grew to include tens of thousands of files and many third-party dependencies. Similarly, the inability to search for symbols in packages forced developers to open external documentation or manually browse site-packages to find function definitions.
These limitations have been a frequent source of user feedback. The March 2026 release directly addresses both issues: the Rust-based parallel indexer removes the performance bottleneck, while the symbol search feature closes a key gap in code navigation.
What This Means
The updates together represent a shift toward a more responsive and seamless developer experience in VS Code. The parallel indexer promises to reduce the time between opening a project and having full IntelliSense from minutes to seconds, making large codebases feel as nimble as small ones.
The ability to search across installed packages means developers can explore unfamiliar libraries — and even jump to internal package code — without leaving the editor. This reduces context switching and keeps the developer in a flow state.
Microsoft encourages all Python users in VS Code to try both features and report feedback. The parallel indexer, particularly, needs broad validation to become the default. If you work on large Python projects, enabling both features now may dramatically improve your daily workflow.