The limitations of models in science are that they don't show you the actual thing, models are very different.
The benefits of using models is the ability to get empirical data for some experiments that can be used to predict full scale results. This can result is faster development or understanding of the experiment underway, and in some cases a safer approach to the intended end product or result.
Another answer: You may mean an intellectual model, a mental picture of how things work. These are tremendously powerful tools in the understanding of science and enable us to make predictions to test their validity, a fundamental part of the scientific method. The only drawback I can think of is that it's easy to become too attached to one, making us blinkered when we find anomalies which should lead to the discarding or updating of the model.