What is Difference Between linear and level option type in SARIMAX prediction function

Question

I am watching Jose Portilla's Time series lectures and in the Sarima video, he says that during prediction we need to use the typ value as "level" instead of "linear" I checked the difference between the two on SARIMAX predict Docs. It says that

‘linear’ : Linear prediction in terms of the differenced endogenous variables.

‘levels’ : Predict the levels of the original endogenous variables.

Can anyone please help me understand the meaning behind this sentence.

I know the level typ get the original value of the predicted output but what about linear typ