Basically a GPU, as that's the highly parallel machine you're most likely to program any time soon. I'm not sure there's anything quite like PRAM in the sense that it's a mathematically abstract model that nonetheless captures something interesting about performance, but Volkov's thesis is a pretty good start for modeling both the memory and ALU cost of a GPU computation:

https://digitalassets.lib.berkeley.edu/etd/ucb/text/Volkov_b...