Would the IP Address: 192.168.0.254 be okay to use for the 7800n?
There's no problem using x.y.z.254 on a /24 (an IPv4 network with a subnet mask of 255.255.255.0) as the only 'special' addresses are the lowest, called the subnet address (x.y.z.0 in this case) and the highest, which is the broadcast address (x.y.z.255 in this case).
It is usual to use either the first address after the subnet address or the last address before the broadcast address for the gateway address. In the case of a /24, this means using x.y.z.1 or x.y.z.254. I use x.y.z.254 on IPv4 /24s, largely out of mostly fond memories of Acorn's Econet (where the default fileserver was station 254 on the local network, expressed as 0.254).
Subnets only seem to be voodoo because decimal numbers are used for IPv4 addresses. The mystery disappears when you use binary. I have a public IPv4 /28. This means the first locally allocated bit in the address is the 28th bit. Subtract 28 from 32 gives you five - so the five least significant (right most) bits are local, making the subnet mask is 27 1s followed by 5 0s (255.255.255.224 in decimal).
My public network is x.y.z.112/28. 112 is 01100000 in binary, so x.y.z.112 is the network address. The broadcast address will be 01111111, which is x.y.z.127 in decimal. The 'normal' addresses in my /28 are x.y.z.113 to x.y.z.126.
Zen use the highest normal address in the block as the gateway address, so my router's WAN interface must be x.y.z.126. I NAT all IPv4 traffic without a dedicated public IP address onto x.y.z.126.
In fact, I use the /28 as two /29s. You can usually get away with using the network address of a subnet smaller than a /24 as a normal address (though I do not do so, as I have enough addresses not to need x.y.z.112 and x.y.z.120).
On IPv6, I follow the usual practice of <prefix>::1 for a statically allocated gateway, though IPv6 routing is typically done using link local addresses in the fe80::/10 subnet advertised, along with the network prefix and prefix length, using Neighbour Discovery.
IPv6 addresses contain 128 bits and are expressed in (up to) 8 groups of four hexadecimal characters separated by colons. An IPv6 subnet is usually at least a /64, though there are some circumstances where smaller allocations are used.
In almost all circumstances, you can omit leading zeroes in each group, also you can omit any number of groups that are entirely zero once only using ::. These rules allow 2a02:0001:03fe:0201:0000:0000:0000:0001 (which I've made up, though it is a globally routable address) to be written as 2a02:1:3fe:201::1. 2a03:2880:fffe:000b:face:b00c:0000:0099, which is a Facebook NS, can be written as 2a03:2880:fffe:b:face:b00c::99.
Typically an ISP gives its customers a /48 (delegation starts with the fourth group) or /56 (delegation starts with the right most two characters of the fourth group). This gives a lot of address space to play with! I have a /48 from Zen, but only use a small fraction of the lowermost /56 (currently the lowermost five /60s). A /56 delegation would be more than ample for my needs and I restrict myself to the lowermost /56 in case Zen ever want to recover the rest of the /48.