Which terminal value method gives a higher valuation - Gordon growth or exit multiple? What does it depend on?

Neither is systematically higher - it depends entirely on the inputs. A high perpetuity growth rate relative to WACC pushes Gordon growth up; a generous exit multiple pushes the exit method up. The honest answer in an interview is that they should produce SIMILAR values once cross-checked, and if they diverge wildly your assumptions are inconsistent. The whole point of running both is to triangulate: each method's output can be reverse-engineered into the other's implied input to test reasonableness.