I would not call Sierpenski's triangle a computational model. A computational model should be a system that has a way to compute something. This means there needs to be a way to supply an input, and a way that the system does something with that input to compute something. Sierpenski's triangle is a single thing (a single set), and it is fixed -- there is no way to supply different inputs and get a different output.
For similar reasons, I would not call Pascal's triangle a computational model.
However, I would call rule 90 a computational model. There are standard ways of providing an input and computing stuff (e.g., the input bitstring is interpreted as a row, and then successive rows are derived using the cellular automaton rules, which is interpreted as a computation). Similarly, I would call rule 110 a computational model. Or, more generally, I might call cellular automata a computational model, and rule 90 and rule 110 as two instances of cellular automata.