Reproducibility is (supposed to be) a cornerstone of science. Model versions are absolutely critical to understand what was actually tested and how to reproduce it.