IPv6 Subnet Calculator

IPv4 subnetting eventually becomes muscle memory — everyone's done a /24 breakdown a hundred times. IPv6 doesn't get the same repetition because the numbers are enormous and the address itself is harder to eyeball, so this does the actual 128-bit arithmetic and hands back the network address, range and address count for any prefix.

The scale is genuinely different, not just bigger

A /64 — the standard subnet size for basically every IPv6 LAN segment — contains 18,446,744,073,709,551,616 addresses. That's not a typo and it's not a design accident: IPv6 assumes you hand out a /64 per subnet regardless of how many hosts are actually on it, because the address space is large enough that "wasting" most of a /64 on a home network or a single VLAN genuinely doesn't matter. This is the single biggest mental adjustment coming from IPv4, where every address was rationed.

Why /64 is the number you'll see everywhere

Stateless Address Autoconfiguration (SLAAC) — the mechanism that lets an IPv6 host generate its own address without DHCP — specifically requires a /64 to work, because the interface identifier occupies the lower 64 bits. Go smaller than /64 on a LAN segment and SLAAC breaks. This is why ISPs typically hand out a /56 or /48 to a site: it's not generosity, it's giving you enough /64s to actually subnet with, the same way a /24 gave you room to VLAN an IPv4 network.

Reading a compressed address correctly

:: collapses one run of consecutive all-zero groups — and only one run, because if it could appear twice the address would be ambiguous about how many zero groups belong to each gap. 2001:db8::1 expands to 2001:0db8:0000:0000:0000:0000:0000:0001. It's worth being able to expand these by hand at least once, because compressed notation is exactly where transcription errors hide — miscounting how many groups :: is standing in for is a genuinely common source of a subnet plan that looks right and isn't.

Where this actually matters day to day

Planning a /48 or /56 allocation from an ISP into individual /64s per VLAN, sanity-checking a firewall rule written against an IPv6 range, or just working out whether two addresses you're looking at are actually in the same subnet — all of it depends on getting this arithmetic right, and doing 128-bit hex math by hand is exactly the kind of thing that's easy to get subtly wrong under time pressure.