For me is because I prefer having a mental model so that I can immediately understand what to expect from a model instead of having to remember every model as a single separate instance.

E.g. I know what to expect from the sonnet vs opus families, so if I read sonnet 4.6 and opus 5.5 I know to which generation they belong and what kind of general capabilities they could offer. if I read opus 9.2 I know something is off. I guess, it helps me organize my thoughts.

With names from Google, if I were to read Gemini Helios, or Gemini Iron I would have no idea how they fit the bigger picture.