As for the number of cycles, in the order of 50 times. limited by the experiments, never seen the semi-rigid failing. (Hence the word probably)
As for Gigalink, mentioned by sixtimesseven, they indeed have some very nice cable. A representative showed me some nice silicone samples, that were supposed to stay flexible at 77K. Never used them.
BTW in general loss does not matter as much at these temperatures, as the added noise due to the loss is only about a quarter of what is would be at room temperature assuming the loss is the same at both temperatures (Bosma's theorem).