Hello,
I currentrly use Substrate storm and I have evaluated SeismIC. You are alright they were "integrated" in SNA, although I dont know any big change in substratestorm that is the software I use most.
I think that they differ in the algorith they use to modelate the subtrate. When I have evaluated SeismIC in the tutorial It surpise me a lot that you only have to describe  the subsrtate doping leval, while in substratestorm you describe the a detailed substrate profile. Then the models extracted by substratestorm are more detailed (this means a large number of Rs and Cs) . I think this is the reson substratestorm is prefered by circuits with few transistors and RF applications. However I have simulated "large" circucits with thousand  of transistors  using the macro capability of SS, Basically It assumes that in a selected area the substrate below is a single node, that is almost always true for Heavly Doped substrates with enogh substrate contacts  but it should be used with care in Lighly doped substrates.
bye