IP addresses explained: IPv4, IPv6 and the reserved ranges

Every device that talks on the internet needs an address that tells packets where to go. That address is the IP address. Most of the time nobody notices it — until something breaks, a firewall rule needs writing, or a website shows an address you do not recognise. This guide covers what the numbers mean, why there are two kinds, and which ones are reserved for special purposes and never appear on the public internet.

IPv4 and IPv6

IPv4 addresses are 32 bits long and written as four decimal numbers from 0 to 255, separated by dots: 203.0.113.10. That allows about 4.3 billion addresses, which sounded generous in 1981 and ran out in practice during the 2010s. The regional registries have handed out their last free blocks, and new networks mostly get IPv4 space by buying it from someone else.

IPv6 addresses are 128 bits long and written as eight groups of four hexadecimal digits, separated by colons: 2001:0db8:0000:0000:0000:0000:0000:0001. Leading zeros in a group can be dropped, and one run of all-zero groups can be replaced by a double colon, so that address is normally written 2001:db8::1. The space is large enough that a single home connection is routinely given more addresses than the entire IPv4 internet.

The two protocols are not compatible on the wire: an IPv4-only device cannot talk directly to an IPv6-only one. So most networks today run both side by side, which is called dual stack.

Why a website sees only one of them

A connection uses one protocol at a time. When a device has both, browsers and operating systems try IPv6 first and fall back to IPv4 only if IPv6 is slow or broken — a technique called Happy Eyeballs. So a website usually sees your IPv6 address if you have one, and your IPv4 address stays invisible to it. Seeing an IPv4 address usually means the connection has no working IPv6.

This is why our IP address tool names the protocol next to the address it shows.

What is my IP?

Public and private addresses

A public address is routable on the internet: a packet sent to it from anywhere in the world arrives at the same network. A private address only means something inside one network. Millions of homes use 192.168.1.1 at the same time without conflict, because those packets never leave the building.

In IPv4, a home router usually hands out private addresses to the devices behind it and translates their traffic to a single public address on the way out. That translation is NAT, network address translation. It is why the address in your computer's network settings — often 192.168.x.x or 10.x.x.x — differs from the one a website sees.

Many mobile providers, and some fixed-line ones, go a step further: they share one public IPv4 address among many customers, using the range 100.64.0.0/10 on their side of the connection. This is carrier-grade NAT. If your router reports an address between 100.64.0.0 and 100.127.255.255 on its internet side, you are behind it — which is why port forwarding may not work, and why a website that blocks or rate-limits one address can end up affecting a whole group of unrelated people.

IPv6 was designed to make NAT unnecessary. Devices usually get public addresses directly, and a firewall, rather than address translation, decides what may reach them. The IPv6 counterpart of the private ranges is the unique local range, fc00::/7, which is used far less than the private IPv4 ranges are.

Reserved IPv4 ranges

These blocks are set aside by the IETF and recorded by IANA. None of them works as an ordinary public address.

RangePurposeWhere you meet it
10.0.0.0/8Private useCompany networks, VPNs, cloud networks.
172.16.0.0/12Private use172.16.0.0 to 172.31.255.255; Docker and some corporate networks.
192.168.0.0/16Private useHome routers: 192.168.0.1 and 192.168.1.1 are the usual gateway addresses.
100.64.0.0/10Shared address spaceThe internet side of a router behind a provider’s carrier-grade NAT.
127.0.0.0/8Loopback127.0.0.1 is the machine itself; the traffic never leaves it.
169.254.0.0/16Link-localSelf-assigned when no DHCP server answered — usually a sign something is wrong.
192.0.2.0/24, 198.51.100.0/24, 203.0.113.0/24DocumentationExamples in manuals and articles, including this one. Safe to publish because they route nowhere.
198.18.0.0/15BenchmarkingTest labs for network equipment.
0.0.0.0/8“This network”0.0.0.0 as “every address” in server settings, or “no address yet”.
224.0.0.0/4MulticastOne sender, many receivers: local streaming, routing protocols.
240.0.0.0/4ReservedNever assigned; most systems refuse to use it.
255.255.255.255/32Limited broadcastEvery device on the local network segment at once.

Reserved and special IPv6 ranges

RangePurposeWhere you meet it
::1/128LoopbackThe machine itself, like 127.0.0.1.
::/128Unspecified“No address yet”, or “every address” in server settings.
::ffff:0:0/96IPv4-mappedAn IPv4 address carried in IPv6 notation, such as ::ffff:203.0.113.10.
64:ff9b::/96IPv4/IPv6 translationIPv6-only networks reaching IPv4 servers through NAT64 — common on mobile.
fe80::/10Link-localEvery IPv6 interface has one; valid only on the local link.
fc00::/7Unique localThe IPv6 counterpart of private ranges; in practice, addresses start with fd.
2001:db8::/32, 3fff::/20DocumentationExamples in manuals. 3fff::/20 was added in 2024 so that examples can show networks larger than 2001:db8::/32 allows.
ff00::/8MulticastIPv6 has no broadcast; multicast takes its place.
2000::/3Global unicastNot reserved: the block ordinary public IPv6 addresses come from.

Reading CIDR notation

The /24 or /64 after an address is its prefix length: how many of the leading bits identify the network. The bits that remain number the hosts inside it. An IPv4 /24 leaves 8 bits, so 256 addresses. Each step down doubles the size: a /23 has 512 and a /16 has 65,536. In IPv6 the usual subnet is a /64, which leaves 64 bits — about 18 quintillion addresses — for the devices on a single network.

CIDR / subnet calculator

What an IP address reveals

An address identifies a network connection, not a person. Everything you connect to sees it — websites, game servers, mail servers — and anyone can look up which organisation it is registered to. It usually reveals the provider and a rough location: often the right country and region, sometimes the right city, rarely more than that. The subscriber behind a connection is known only to the provider, which normally discloses it only when legally required to.

Because the address belongs to the connection, it changes when you change networks — home Wi-Fi, mobile data, the office — and many providers also reassign it from time to time. IPv6 devices add a rotation of their own: temporary privacy addresses change daily or more often, so that a single address does not follow a device around for months.

WHOIS / RDAP lookup

Frequently asked questions

Is 192.168.1.1 my IP address?

It is the address of your router on your own network, and the same address is in use in millions of other homes. The address the internet sees is your public one, which is what the IP address tool shows.

Why does a website show an IPv6 address when I expected an IPv4 one?

Because your connection has both and the browser preferred IPv6. Your IPv4 address still exists; it just is not the one this connection used.

Can two devices have the same IP address?

Not on the same network. On different private networks they can and constantly do, and behind NAT many devices share one public address.

What is the difference between a static and a dynamic IP address?

A static address stays the same until you or your provider change it. A dynamic one is assigned from a pool and may change when the connection restarts. Most home connections are dynamic; servers and some business lines are static.