The phrase "order of magnitude" is used to indicate differences between quantities in terms of exponential powers. I've also seen it to indicate Big-Oh differences in algorithm run times.
However, in computer science there is also a notion of classifying problems in terms of their qualitative difficulty. This is the categorization of P, NP, NP Hard, NP Complete, etc. I don't think saying that an NP problem is an "order of magnitude" harder than a P problem is well-defined. Is there a different phrase which suggests the qualitative jump in problem difficulty?