Skip to content

Third-party tools

This application includes integrations with multiple third party tools; these are not installed by default, but may be installed from the "Tools" window, accessible from the GUI. These are called as either standalone executables, or processes managed by a dedicated python interpreter and environment, depending on the process.

For processes which use a dedicated python interpreter, Molchanica handles setup and management for this automatically, and uses uv to set up isolated python environments at the correct version, with the correct dependencies.

The image below shows the tool window, and lists the tools available as of 2026-08-03. Some applications like ORCA must be manually installed due to license or other considerations.

You can click the ? button by each tool to show details, including its version, license, and directory it's installed to.

Third-party tool window

Where tools are installed to

Tools are installed to process_executables/, inside the one folder Molchanica writes to: the folder its executable is in. Python-backed tools each get their own isolated environment, under process_executables/python_envs/<tool>. See the data folder section for the full layout, and for where that folder is on each platform.

The Tools panel has a button that opens process_executables/ in your file browser; the ? button by each tool shows the exact directory that tool is installed to.

Earlier versions kept these tools in the operating system's per-user data directory instead — %LOCALAPPDATA%\molchanica\ on Windows, $XDG_DATA_HOME/molchanica/ or ~/.local/share/molchanica/ on Linux. Molchanica moves them for you on first launch after upgrading. This is instantaneous when both locations are on the same drive, which is the normal case. If they're on different drives, the move is skipped rather than spending a long time copying gigabytes at startup; Molchanica prints where the old files are, and you can move them yourself.

Warning: This directory can get very big, e.g. if you install multiple tools. Many of them include large ML models and frameworks, and independent CUDA libraries.