Representative gene models

[Guest]

From a practical standpoint, what would be the difference between using all gene models and representative gene models from a genome project for e.g. annotation of another genome?

I often see that in addition to listing all gene models there is also another collection of just representative gene models which is offered (list is smaller in size). From what I understood is that in the representative gene models collection only one (the longest) splice variant is given.
When would be it be advisable to use one or the other collection?

In this example I intend to use it for genome annotation so going with the (larger) collection of all models seems to be intuitive, but I'm just wondering about other uses...

Thanks!