The possible areas of growth within them are; speed, differentiating hundreds and thousands places (when dealing with money), flipping of numbers (50 vs 05), attention to detail of operation (+ vs -), and accuracy in calculation. He will also benefit from strategies to organize his work, as well as developing his ability to show his thinking in a written (non-verbal) format.”
Like Malraux, I'd be a bit concerned that all these are processing skills, not conceptual challenge. These skills will almost certainly improve with age; they may or may not improve a worthwhile amount with excessive practice while he is still this young.
From my own experience, I can say that nothing will teach a math-loving child to hate math faster than forcing them to spend all their time on basic computation, processing skills and writing. (Multiply exponentially when the kid has challenges with fine motor, writing, processing speed, attention, etc). These may be challenge areas, and addressing them is important, but doing so as a primary focus can be detrimental. We've found it really important to allow our math monster to explore and be challenged with new concepts and more complex problem solving. When school focused only on the kinds of skills you list above, he became anxious and withdrawn. Eventually, he decided that not only did he hate math, he was also no good at it: it demanded only his weaknesses, and made no use of his conceptual math strengths.
So long way of saying - yes, of course it's important to work on the processing bits. But not to the exclusion of growth on the conceptual side. For us, those skills were far more readily improved when they were practiced in the context of a motivated kid engaged in challenging problem solving. (Thank god for AoPS! He learned to love math again.) So personally, I would look at which option will more likely be able to provide conceptual challenge, and scaffold the other parts as needed while he builds up skill in the mechanics.
It seems like you may be in one of those rare situations where differentiation would actually, really happen, so the common concerns of implementation, in what ways and how often, and whether actual different instruction would be involved seem less concerning. The other big consideration may also not apply here: that is, if he gets differentiation instead of skipping, what is the likelihood next year's teacher may not recognize that work, and make him repeat the same stuff in the next grade?
