@GordonFreeman
Currently we have four full length official GMAT computer adaptive tests that are as close to the real exam that we can get. Two come from the free GMATPrep Practice tests, and the other two from the Exam Pack 1 extension. I think that is plenty in my opinion, and using one of them to get an initial baseline is not a bad idea. There is no sense in saving them for the end because the whole idea is to get a good sense of what the real exam will be like. For example, when you take one of the GMATPrep tests, it really demonstrates how difficult the real GMAT will be like. In my opinion, it is better to get that experience early on in the preparation, so one can plan accordingly.
In your case, because you are a strong student, it is okay to take the GMATPrep early on, it will be stressful but you will do just fine. Of course, after you prepare for a month or so, your score will go up. In my experience working with students, they typically improve about a 100 points from when they take the GMATPrep initially. To give you an example, I had a very strong student who scored a 630 when he first took the GMATPrep cold. To many that may not seem like a high score, but it is actually a very good score when you are taking it cold for the first time. Taking one of the GMATPrep tests is very stressful and really puts a lot of pressure on students. He scored a 99% on the Verbal, which is very impressive, and a lower score on the quant because he didn't complete the section. So things like that happen. Eventually, he scored a 730 after a month of preparation. I would not recommend you to take the GMATPrep test if you hadn't done as well on the diagnostic test, because in that case GMATPrep would cause more psychological damage than help. This is just my opinion based on experience working with students having a range of abilities.
There are several issues I have with the non-Official GMAT CATs. The first is that the scores are not reliable, so if someone gets a 680, then we are not really sure if that translates to 680 on the real exam. You will see a lot of threads in this forum on this issue. The second is that the style, format, and difficulty level of the real GMAT is almost never captured by any of the commercially available CATs. Some companies go overboard in complicating the problems, to make them seem hard. Others will often have mistakes or ambiguous statements. To be honest none of the companies put in the amount of effort and resources that the test writers do in creating the questions. For example, all of the official GMAT questions are first tested for ambiguity and bias by using them as part of the experimental questions that are administered in the real GMAT.
So that in a nutshell is the logic of recommending official GMAT questions. This will make more sense as you continue your GMAT preparation.
Cheers,
Dabral