Thanks😊 It's primarily prompt driven. The same model is used across all modes, but each explanation style has its own carefully designed prompt that guides the depth, terminology, and tone of the response. I experimented with several prompt strategies before settling on the current approach to get more consistent results.