A 37 pt difference between VCI and WMI would certainly make the FSIQ suspect. Not uninterpretable, precisely, but to be viewed with significant caution. This may or may not be an appropriate situation for the GAI, partly because I believe you indicated that his PSI is substantially higher than his PRI. (Which is probably why the GAI is only 8 points higher than the FSIQ--significant, and rare enough in the general population, but not remarkable (20% base rate) in the GT population.) The best case for the GAI is when both VCI and PRI are high, and both WMI and PSI are low. Second to that would be VCI and PRI high, and one of WMI or PSI is somewhat high, and the other is much lower. I do lean toward using the GAI in the GT population, in most cases.
His PRI subtest scores suggest that he has exceptional abstract thinking skills (matrix reasoning), but only average concrete nonverbal skills, including perceptual-motor (block design). The latter may be affected, of course, by visual and fine-motor skills, in addition to actual visual-spatial problem-solving skills.
Math is about logic, patterns, and reasoning, which is also what matrix reasoning is about. Block design is primarily perceptual in nature. And it's timed, which is not good for those who like to sit and ponder deeply. Picture concepts is concrete, and relies to some extent on verbal concepts. So it is significant subtest scatter, but it's not that strange that a math-y kid would score exceptionally well on mr, the abstract task, but nothing special on bd and pc, the more concrete tasks.
