Subnet calculator:
the mask is the seam.

Every IPv4 subnet spends its lowest address as the network identifier: all host bits zero, the name routing tables use. It spends its highest as the broadcast: all host bits one, which addresses everyone in the block at once. Neither can be assigned to a device, so a /24's 256 addresses yield 254 usable. Two exceptions are worth knowing. A /31 point-to-point link (RFC 3021) has no broadcast and uses both addresses; a /32 is a single-host route. IPv6 dropped the whole scheme — there is no broadcast address to subtract.

No. RFC 1918 defines exactly three private ranges: 10.0.0.0/8, 172.16.0.0/12 and 192.168.0.0/16. The /12 is the trap: it covers 172.16.0.0 through 172.31.255.255 only. 172.32.x.x is public address space that belongs to someone else, and traffic you meant to keep internal will eventually route toward them. The calculator prints the scope for every address, so this shows up before deployment rather than after.

A wildcard mask is the subnet mask with every bit flipped: a /26's mask 255.255.255.192 becomes wildcard 0.0.0.63. Cisco access-control lists and OSPF network statements use this inverted spelling: 0 means "this bit must match", 1 means "don't care". It describes the same boundary as the mask, from the opposite direction. We print both, and you can paste either into the box above.

Variable-Length Subnet Masking gives each network the smallest block that fits, instead of splitting the parent into equal pieces. A 500-host department gets a /23, a point-to-point link gets a /30, and both live in the same /22. The largest-first rule is about alignment: a block can only start on a multiple of its own size (a /23 on a 512 boundary, a /24 on 256). Allocate big to small and every block lands exactly where the previous one ended, gap-free. Allocate small to big and the later large blocks must skip ahead to their next legal boundary, stranding addresses in between. The planner does the sorting and the alignment and shows what is left over.

Because SLAAC is specified to work with a 64-bit interface identifier, full stop. SLAAC is the mechanism that lets IPv6 hosts configure their own addresses without a DHCP server. Make the subnet smaller and autoconfiguration breaks. So the practical unit of IPv6 planning is the /64 LAN. An allocation is measured in how many /64s it cuts into: a /56 gives 256, a /48 gives 65,536. The exceptions are deliberate: /127s on router-to-router links (RFC 6164) and /128 loopbacks — places where no host will ever autoconfigure.

Not to any router built in the last three decades. Classes (A/B/C by the first bits of the address) were how space was allocated before 1993, and they wasted it badly. The smallest classful unit an organisation could get was a /24, and the next step up was a /16. CIDR replaced the whole scheme with explicit prefix lengths, which is what every routing table uses today. We still print the class because certification exams, interview questions and thirty-year-old configs keep the vocabulary alive — but nothing about how your traffic moves depends on it.

Nowhere. The parser and all the maths run as JavaScript in your browser, so calculating makes zero network requests. Confirm it in your browser's network tab. Your internal addressing is exactly what should not be pasted into a random web tool, so this one never receives it. What you type is never written into the URL either, because a URL carrying your addressing would reach our access log through the Referer header the moment you clicked a link. And there is no JSON API, because an endpoint would be a standing invitation to send it to us. The shareable link under each result is assembled locally and only travels if you copy it.