There are not specific rules governing which rank of higher classification a group belongs to, so different references will define the same group as a class or a subclass, as you've noted. Mostly people just follow one of three standard references for higher classification of fishes -- Eschmeyer, Nelson, or FishBase. All of these differ as follows.
In Eschmeyer there are four living classes of bony fishes: Cladistii (bichirs), Actinopteri (all ray-finned fishes except bichirs), Coelacanthi (coelacanth), and Dipneusti (lungfish). This implies that Osteichthyes would be a superclass or subphylum, and the various tetrapod groups would be classes as well. The FishBase classification is similar except that Actinopteri is further subdivided into three separate classes for sturgeons, gars/bowfins, and teleosts.
In Nelson Osteichthyes is a class, Actinopterygii (ray-finned fishes) and Sarcopterygii (coelacanth + lungfish + tetrapods) are subclasses.
In general, the Eschmeyer/FishBase approach is optimized for living species and Nelson approach is optimized for fossils, as it helps with the multitude of ancient extinct fish lineages.