That's tricky! It would vary significantly by author and (sub)theory. "Generativism" is VERY broad, but even something like "Government and Binding" or "Minimalism" will have some variation.
Another problem is the simple use of the verb "have"-- what does it mean for a sentence to "have a vP"? At what level of representation? Must that vP have a head? Can it be omitted? And if it has a head, can that head be null?
And if does "have a vP" then must we always represent it in the tree? Often various irrelevant (null?) phrases are omitted from trees and analyses. Is that just to save space/ink, or is that theoreticall motivated?
In the end, there are two (extreme) cases to consider:
1. Structure is built as needed. Extra categories are not included except where justified.
So here, no, a vP is only used when it is really needed to represent something.
This would allow language-specific structures!
2. All language have the same basic skeleton for sentences, or at least within a given language (eg, English) all sentences have the same structure. All categories are included in every sentence, always (like TopicP and other CP-domain phrases). Of course we might skip over writing them because they're implied, but they're there.
This would suggest not only that there is some rigid system for English sentences but that the categories have real meaning (cognitively, possibly encoded genetically in humans) and probably suggest that there could be universal tree structures.
Some work by Kayne, for example, seems to assume such universal, very large, structures.
Obviously in the case of vP specifically it might fall somewhere between those extremes.
Based on why you're asking:
If this is for a course: ask your instructor! (Either answer is fine in general, but one will get you the right grade.)
If this is for your own research or general studies: both could be argued, so it's worth comparing them.
