I just tested and PyStan works in Google Colab, which allows running Python notebooks in your browser. This can be useful for demoing or when giving a course and someone has problems with C++ installation in their laptop.
PyStan is not installed by default, but running
!pip install pystan
works, and models compile successfully (maybe slower than in modern laptop, but not too bad).
You can open notebooks also directly from github. In addition of the notebook you need upload Stan model files and possible utility functions (like Mike’s diagnostic utilities) , but again for demos and courses this works quite easily.
If one is going to do case studies so that the idea is to run them on colab, then one could precompile + pickle + compress + “filename with platform and version info” + small function to load the models or recompile them … just saying.