First note that there are two ways a provider can assign you IP addresses (this applies to both v4 and v6 but of course getting a big block of v4 addresses out of your provider is far more difficult than getting a big block of v6 addresses out of them).
Firstly they can route them as "on-link" addresses on the network your server is connected to. If they do this then your server will need to respond to ndp (arp for ipv4) requests for all the addresses you want to receive traffic for. Linux will do this by default for addresses that are added as "local" but if you want to use your machine as a router (for VPN connections or VMs or whatever) then you have to get into the messy buisness of proxy NDP (proxy ARP for ipv4).
Alternatively they can route them to your server's primary IP. In this case your server only has to answer NDP (arp for IPv4) for it's primary IP.
From the looks of linode's documentation it looks like they do the former for their /112 blocks and the latter for their /64 blocks.
So getting back to the question of how to use them, that rather depends on what you want to use them for.
The intent of giving you as a customer a /64 or /56 is not that you will use every sigle address, it is that it is strongly encouraged that all IPv6 subnets should be /64. The original reason for this was to support stateless auto-configuration but it became a broader convention even in scenarios where stateless auto-configuration is stupid.
To the best of my knowledge there is no way to configure Linux to treat a block of addresses as "belonging to the local machine" without adding them individually (which is obviously impractical for a /64). I guess you could write a program that watched a TUN interface and added IPs as local on-demand but it would be setting yourself up for a DDOS.
/24IPv4 subnet, do you assign all 254 addresses to one machine?