Real protocol stacks do not implement seven separate layers, and several protocols cross the boundaries. Given that, why is the model still worth keeping?
Answer
Because it narrows a problem down to which function failed, what information is available there, and which component handles it
The model earns its place as a diagnostic and vocabulary tool, not as a picture of any real stack. Vendors document against nothing in particular and are under no such obligation. The claim that it describes the Internet’s actual architecture belongs to the TCP/IP model, which is the four-layer view; OSI is explicitly a reference model rather than the protocol suite that won. And there is no per-layer tool that isolates one layer cleanly — a single capture shows several layers at once, which is part of why you need a way to decide which one you are asking about.