What is a strategy and what is a model depends mostly on the words the author likes to use. In this case you could substitute one for the other.
Best effort is essentially no QoS. You might say the strategy is "no strategy," or the model is "no model."
DiffServ is the most common QoS model (or strategy, whatever). Data classes are predefined, and bandwidth is allocated per class on each device. The number of classes has increased over time as switch hardware capabilities have improved, and as applications that require specialized handling (realtime audio/video) have become more common.
Sometimes, ideas that look great in the lab don't work out in practice. IntServ was one such idea. The thought was that hosts wold tell the network how much bandwidth they needed, or how much latency they could tolerate, and the network would configure itself to provide it. Each router along the data path would "reserve" bandwidth when requested. When the data flow was finished, the router would release the reservation.
It was a great idea, but in practice it didn't work out so well. The biggest problem was that each router had to maintain the state of every data flow. This put a tremendous burden on memory. Also since the send and receive paths could be asymmetric, the signaling messages didn't reach the right devices.
IntServ never really caught on, and DiffServ has become the dominant QoS strategy (oops, I mean model).