Comment by Turskarama

Both the context and the prompt are just part of the same input. To the model there is no difference, the only difference is the way the user feeds that input to the model. You could in theory feed the context into the model as one huge prompt.