IPv6 Subnet Calculator

How IPv6 subnetting works

Where IPv4 had 32 bits, IPv6 works with 128-bit addresses, which works out to 2¹²⁸ ≈ 3.4 × 10³⁸ unique addresses. You write one as 8 groups of 4 hexadecimal digits joined by colons (say, 2001:0db8:85a3:0000:0000:8a2e:0370:7334), and the :: shorthand collapses one run of consecutive zero groups down to nothing (2001:db8::1). The prefix tells you how many bits belong to the network. A /64 is the standard for any LAN, giving 2⁶⁴ ≈ 18.4 quintillion hosts per subnet and making SLAAC possible. A /48 is allocated per customer site, which is 65,536 /64 subnets, while a /56 is common in residential broadband. The host count in a prefix /n is 2^(128−n), and the number of /64 subnets inside a /n is 2^(64−n). So a /48 holds 2⁽⁶⁴⁻⁴⁸⁾ = 65,536 /64 subnets. RFC 4291 reserves the 2000::/3 range for global addressing.

Applications

ISPs lean on this when they plan customer allocation, delegating a /56 or /48 over DHCPv6-PD. Datacenters use it to segment VLANs across /64 racks. In IoT, every device can take a unique global address with no NAT in the way. Corporate networks route between sites with a /48 per location. It also covers ARIN, RIPE and LACNIC allocations, staying compliant with RFC 4291 and RFC 6177, and migrating off IPv4 using dual stack.

FAQ

Why is /64 the standard for LAN? SLAAC (RFC 4862) and EUI-64 build host addresses from the MAC using the lower 64 bits. Carve a subnet smaller than /64 and SLAAC stops working.

Are there reserved or unusable addresses inside the subnet? Basically none. IPv6 dropped broadcast and never adopted IPv4's host-bits convention. There's the all-zeros address (subnet-router anycast) and a handful of special suffixes, but they don't really dent your host capacity.

How much does a typical home receive? Most residential broadband subscribers are handed a /56 or /60 through DHCPv6 Prefix Delegation. That's room for 256 or 16 internal /64 subnets, respectively.