Wouldn't think SiC diodes would have any advantage... Rs is much higher and you don't get any advantage from Vrm. Unless you somehow arranged it so you could deliver 500V into the LO port without it being shorted away by two Vfs, which would give pretty ridiculous dynamic range I suppose. But given the sizes commercially available, it would only work up to HF or so.
A diode-strapped BJT might be of interest. Relatively low capacitance and quite fast switching (much faster than a single junction alone -- effectively, self Baker clamped). 2N3904 is supposed to be nanoseconds fast, though Cbe isn't going to help much relative to, say, 1N914. A particularly small type might fare better (but beware of the low Vebo of RF types).
Tim