I think it's useful to understand how git actually stores objects as blobs and trees, but the mental model of "commits are diffs" is still useful. Commands like git-cherry-pick really make the most sense when thought of in terms of diffs. Internally, for a command like that, git is taking the diff and applying it, anyway, so the model isn't really that inaccurate.

I think the main place I've seen where the mental model of "commits are diffs" breaks down and actually leads to significant confusion and/or mistakes, is when the tree ends up in a weird state, such as in the middle of resolving a merge conflict.

In that situation, it's important to keep in mind there's three separate tree-like things: the actual working tree; the index; and HEAD. And git status shows nothing more or less than the diff between the working tree and the index. I've gotten myself out of some confusing situations just by reminding myself of that and thinking things through very carefully.