Enter to send·Shift+Enter new line
wrap_model_call( self, request: ModelRequest, handler: Callable[[ModelRequest], ModelResponse] )
Raise when the reply was cut off before any output.
