Find your IP

IPv4 vs IPv6: what is the difference?

IPv4 is the internet's original addressing scheme and IPv6 is its replacement, and in 2026 you are almost certainly using both. This guide explains how the addresses differ, why the change was needed, whether IPv6 is faster or safer, and how to see which one your connection uses.

In this guide
Short answer

IPv4 uses 32-bit addresses written as four numbers (203.0.113.42), about 4.3 billion in total, which ran out and are now shared through NAT. IPv6 uses 128-bit addresses written in hexadecimal (2001:db8::1), enough for every device to have its own public address, with a simpler header and no need for NAT. The two do not interoperate, so the internet runs both side by side; your browser prefers IPv6 when a site offers it. Neither is meaningfully faster or safer by itself: speed depends on the path, and security on HTTPS, not on the address family.

Address format and size

An IPv4 address is 32 bits long, written as four decimal numbers from 0 to 255 separated by dots. The leading part of the address names the network and the rest names the device within it; a prefix length such as /24 says where the split is. 32 bits give 232 = 4,294,967,296 addresses, and a sizeable share of those is reserved for private networks, multicast and other special purposes, so the usable public pool is smaller still.

An IPv6 address (RFC 4291) is 128 bits, written as eight groups of four hexadecimal digits separated by colons. Two shorthand rules keep it readable: leading zeros in a group are dropped, and one run of consecutive all-zero groups can be replaced by ::, so 2001:0db8:0000:0000:0000:0000:0000:0001 is written 2001:db8::1. In practice a home or office network is a /64: the first 64 bits identify the network and the last 64 identify the device, which leaves 18 quintillion addresses per network. Providers typically assign a /56 (256 networks) or a /48 (65,536 networks) to a customer. The subnet calculator expands, compresses and splits either kind.

Anatomy of an IPv4 address (32 bits in four 8-bit octets, written 203.0.113.42) compared with an IPv6 address (128 bits in eight 16-bit groups, written 2001:db8:85a3::8a2e:370:7334, with a 48-bit routing prefix, a 16-bit subnet ID and a 64-bit interface identifier). IPv4 — 32 bits 203011342 8 bits8 bits8 bits8 bits network part (/24 here) · host part · total 4,294,967,296 addresses IPv6 — 128 bits 2001:0db8:85a300000000:8a2e:0370:7334 global routing prefix · 48 bitssubnet · 16interface identifier · 64 bits written 2001:db8:85a3::8a2e:370:7334 — leading zeros dropped, one run of zero groups shortened to :: eight 16-bit groups · 340,282,366,920,938,463,463,374,607,431,768,211,456 addresses (2^128)
IPv4 packs four 8-bit numbers into 32 bits; IPv6 has four times as many bits, with the last 64 normally identifying the device inside a /64 network.

Why we ran out of IPv4

Four billion addresses sounded limitless in 1981, when the IPv4 specification was published for a research network of a few hundred machines. Early allocations were generous (whole /8 blocks of 16 million addresses went to single universities and companies), the classful scheme wasted space between sizes, and then the web, broadband and mobile phones each multiplied the number of connected devices. IANA handed out its last five /8 blocks to the regional registries on 3 February 2011; APNIC reached its final allocation policy that April, RIPE NCC in September 2012, ARIN in September 2015, and the remaining registries by 2020. Since then, new IPv4 space comes from returns, transfers between organizations and a secondary market where a single address costs tens of dollars.

Three things kept IPv4 alive anyway. Classless routing (CIDR, 1993) let blocks be sized to need. Network address translation let one public address serve a whole household. And carrier-grade NAT let one public address serve many households, at the cost of broken peer-to-peer connections, port forwarding and some games. The long-term answer, IPv6, was standardized in 1998 and finalized as RFC 8200 in 2017; the delay in adopting it is a story about incentives, not technology.

Dual-stack

IPv4 and IPv6 packets are different formats and a device speaking only one cannot talk to a device speaking only the other. The transition strategy is therefore to run both at once, "dual stack": your device gets an address of each kind, a site publishes both an A record (IPv4) and an AAAA record (IPv6) in DNS, and the browser tries both, preferring IPv6 but falling back within a few hundred milliseconds if it does not answer (the "Happy Eyeballs" algorithm, RFC 8305). Where a network is IPv6-only, mostly mobile carriers, NAT64 and DNS64 translate on the fly so that IPv4-only sites still load. Google measures the share of its visitors arriving over IPv6; in March 2026 that figure crossed 50 % for the first time, with France and India above 70 %, Saudi Arabia at 65 %, and some countries still in single digits.

NAT and CGNAT

With IPv4, your router translates every device's private address to the one public address it was given (NAT, RFC 3022); with carrier-grade NAT (RFC 6598) your provider does it again for a group of customers. Each layer breaks the original assumption that any device can reach any other: incoming connections need port forwarding, which CGNAT makes impossible, and some games and calling apps need workarounds. IPv6 removes the need: every device has a public address, and the router's firewall, not address shortage, decides what may come in. If your router's WAN address is in 100.64.0.0/10, you are behind CGNAT, and IPv6 is the clean way out; what is CGNAT explains the symptoms.

Is IPv6 faster or safer?

Speed. The protocol itself is not faster: packets travel the same fiber at the same speed. IPv6 has a simpler fixed-size header (40 bytes, no checksum, no fragmentation by routers), which makes forwarding slightly easier, and connections that skip a CGNAT layer avoid its translation step, so large content networks have reported IPv6 slightly ahead on mobile networks. Equally, a poorly peered IPv6 path can be slower than the IPv4 one. The difference is usually a few milliseconds either way; what you notice as speed comes from your provider's capacity and the distance to the server. Measure yours with the ping test over each family rather than assuming.

Security. IPv6 was designed with IPsec in mind, and it is often said to be "more secure"; in practice IPsec is optional on both families and almost nothing on the web uses it either way. What protects your traffic is HTTPS, which works identically over both. Two real differences: the IPv6 address space is far too large to scan the way IPv4 is scanned, and there is no NAT to accidentally hide you, so your router's firewall must do that job explicitly (every consumer router blocks incoming IPv6 by default). One privacy point is worth knowing and is covered next.

IPv6 privacy addresses

The last 64 bits of an IPv6 address used to be derived from the device's MAC address, which would have made every device trackable across networks forever. Modern systems instead generate temporary addresses that rotate (daily by default) for outgoing connections (RFC 8981), and a stable address per network that is not derived from the hardware (RFC 7217). The network prefix still identifies your home, as the IPv4 address does; the device part no longer identifies the device. If you use a VPN, note that a VPN carrying only IPv4 leaves your IPv6 traffic outside the tunnel: the VPN test checks for exactly that leak.

How to check if you have IPv6

The card below reports whether this page reached our servers over IPv6, IPv4 or both, and which one your browser preferred. The IPv6 test adds the detail: your address of each family, the network each belongs to, and whether your DNS resolver answers over IPv6. If you have no IPv6, the reason is usually one of three: your provider has not deployed it (ask, or check its support pages), your router has it switched off (look for IPv6 under WAN or Internet settings, typically "Automatic" or "DHCPv6 with prefix delegation"), or a VPN is in the way, since many VPN apps disable IPv6 rather than tunnel it.

On the device, Windows lists an IPv6 Address in ipconfig (a global one starts with 2 or 3; fe80:: is link-local and means nothing about internet access), macOS shows it under the connection's details, and ip -6 addr prints it on Linux. How to find your IP address has the exact paths. There is nothing to configure on the device itself: IPv6 is on by default in every current operating system.

Test it now

Checking
IP websites see216.73.216.52
IPv62001:db8::1

Running against our own test servers. Nothing is stored.

Open the IPv6 test

Questions

Which is better, IPv4 or IPv6?

IPv6 is the better-designed protocol and the only one with room to grow; IPv4 is the one every service still supports. For a home user there is nothing to choose: leave both enabled and the device uses IPv6 where it is available and IPv4 elsewhere. Turning IPv6 off only costs you the direct, un-NATed path.

Is IPv6 worth it?

For a provider or a company, yes: it is the only way to add customers without buying scarce IPv4 addresses, and it removes the cost and breakage of carrier-grade NAT. For you it costs nothing and brings a public address per device, better peer-to-peer connectivity for games and calls, and no CGNAT problems. It is already on for about half of internet users.

What devices use IPv6?

Every current phone, computer, tablet, console and smart TV supports it, and most mobile networks in large markets run it by default (several are IPv6-only internally). Whether a given connection uses it depends on the provider and the router, not on the device.

What is IPv4 used for?

Everything that IPv6 is used for, on the networks and services that have not deployed IPv6 yet: the two carry the same web, mail, video and gaming traffic. IPv4 remains necessary because a large minority of sites and networks are still IPv4-only, which is why devices keep both.

Can two devices have the same IP address?

Not on the same network: the conflict breaks both. Across different networks, yes: millions of homes use 192.168.1.1 for their router, because private addresses are only meaningful locally. On the public internet, NAT lets many devices share one public IPv4 address, and IPv6 gives each its own.

Sources and further reading

  1. RFC 791 — Internet Protocol (1981)
  2. RFC 8200 — Internet Protocol, Version 6 (IPv6) Specification
  3. RFC 4291 — IP Version 6 Addressing Architecture
  4. RFC 8305 — Happy Eyeballs Version 2: Better Connectivity Using Concurrency
  5. RFC 8981 — Temporary Address Extensions for Stateless Address Autoconfiguration in IPv6
  6. RFC 7217 — Stable and Opaque IIDs with IPv6 SLAAC
  7. RFC 3022 — Traditional IP Network Address Translator
  8. RFC 6598 — Shared Address Space for CGNAT
  9. IANA — IPv4 address space exhaustion (final /8 allocations, 3 February 2011)
  10. Google — IPv6 adoption statistics
  11. Internet Society Pulse — IPv6 reaches a majority of Google's users (March 2026)
CheckMyIP Team
Network engineers and privacy researchers