Frankly, for stuff like hand poses or other stationary poses that are easy to do in front of a mirror, I do self-reference a lot xD
buuuuuuut I still think it's a valuable skill to be able to parse and break down a reference that's not exactly what you need, and learn how to turn it into what you need. In the long run it helps to reinforce a stronger understanding of how objects work in space & will save a lot of time trying to track down or make a perfect reference every time.
3D models are a slippery slope imo. They can be an incredible tool that can potentially be loaded with information, but have a few potential pitfalls too. Setting them up to be correct takes time- For something like a key location in a story that will get used over and over again this could definitely be a time saver in the long run, but trying to model every single one-off location rather that just learning how to draw things in perspective, or modeling a character in every single panel rather than just learning anatomy and how to build poses from scratch or reference has the potential to be a huge time sink. The other biggie is that overuse of 3D models runs the risk of making environments look too "perfect" and tidy, or characters look too stiff and awkward.
Rarely I do use 3D models if I'm having a heck of a time getting some foreshortening right in an extreme angle shot. For example, I don't have a picture of it handy but a few months back I was working on a panel that a group of 5 characters, weapons drawn, running towards a cave. The "camera" was positioned overhead and behind them all, and man- I could NOT figure out how some of their legs were supposed to look at that angle, so I did bust out some 3D models in CSP for that particular panel