It's interesting that you ask how they model the long dated volatility, but not the front month. Why? What, in your mind, makes front month volatility different?
In simple terms, they model the volatility surface and then calibrate it based on the market prices.