Anoplotherium was a tylopod, a member of the clade of Artiodactyls that includes the modern camels. The animal is known from good fossils, so this reconstruction may be very close to the real thing: a generalized, camel-like forager.

This Palaeotherium, like many early reconstructions, show it as a tapir-like animal. Real palaeotheres were indeed perissodactyls, but more closely related to horses and their ancestors than to tapirs. So the actual animal would've probably lacked the proboscis. The tail could be longer and the body less hunched. Imagine Hyracotherium. That's what Palaeotherium probably resembled closely.

Megaloceros may not have looked exactly like an elk (Cervus) with big antlers, as seen here, but this is very close to the real animal.

The Crystal Palace Cenozoic mammals are much better than the Mesozoic reptiles, that's for sure. Not much is wrong with it. All the iconic features are there. Though the tapir-like proboscis is an interesting addition. I'm not sure if real giant ground sloths had one.

Dicynodon was a... well, a dicynodont: a specific group of protomammalian synapsids (otherwise known as the coolest animals, but don't tell the other mods!) that has sharp beaks and two long, protruding canines. Hence the name dicynodont or 'two dog-like teeth'. These two Dicynodon look like turtles! Nothing like the real animal. If you asked me to identity it without clues, I couldn't tell you what it was supposed to be! A real Dicynodon was a burrowing animal with a short tail, sprawling legs with clawed digits, a boxy head with a sharp beak and teeth. It could've had fur like modern mammals, but that remains to be seen until appropriate fossil evidence has been found.

The term 'labyrinthodont' is no longer used today. It was a catch-all term for temnospondyls and similar 'amphibians' with conical, infolded teeth. So with that in mind, there is no basic animal I can really compare this to, so I'll go with Eryops, which is known from fossils in Europe. These models are made to look like toads, complete with bumpy skin and a tympanum. Eryops itself looks very different from a typical toad, and acted different too. It was a large, mostly terrestrial, scaly animal that patrolled the Permian ecosystem. The tail would've been much longer. The body would've been less arched and more streamlined in form. The head, while pretty big, was flattened with eyes facing upwards in the skull.

Quite a makeover. Extreme Makeover: Euryapsid Edition. Modern ichthyosaurs look very much like dolphins, in that their streamlined, fully aquatic animals with dorsal fins, tail fins, and paired foreflippers for swimming. The eye wouldn't even have looked so big and the sclerotic ring would've been covered by the eye-tissue in life. I've seen images of Owen's ichthyosaurs, and they don't have tail fins, so that needs to be fixed.

Much like the Mosasaurus, Plesiosaurus more than likely could not crawl onto land either. The actual animal itself is not known from very good remains and has sort of become a wastebucket taxon for different plesiosaurs, so I'll have to be general with the critique. Basic physics tells us that the necks of plesiosaurs were too heavy to be lifted up like that, and were much more movable under the water, where it could be used to catch fish and squid. Beyond that, the body itself was probably not so stretched out and would've been more streamlined. Teleosaurus... actually looks very close to the real animal, although the snout wouldn't have been so long and gharial-like.