Gemini 3.8 Speed

So far, I find the 3.8 model within Google AI Studio to take much longer on requests⚊upwards of 10+ minutes to complete a task. Do we think this is because it is more advanced and therefore taking more time to be thorough or something else?