The model answers with text.
the tokens the provider reports for this answer, to test a count or a budget.
true for an answer that was cut off at the model's output limit.
true