IPv4 vs IPv6 — what is the difference?

IPv4 and IPv6 are the two versions of the Internet Protocol — the addressing system that gives every device on the internet a number. They do the same fundamental job, but they differ in one decisive way: how many addresses exist. IPv4, the original, offers about 4.3 billion addresses and famously ran out. IPv6, the successor, offers around 3.4 × 10³⁸ — an effectively unlimited supply. Today they run side by side, and most of the internet quietly speaks both.

The core difference: address length

Everything about the two versions follows from the size of an address.

• IPv4 uses 32-bit addresses. That gives 2³² ≈ 4,294,967,296 possible values — roughly 4.3 billion. They are written in dotted-decimal notation: four numbers from 0 to 255, separated by dots, like 203.0.113.5.
• IPv6 uses 128-bit addresses. That gives 2¹²⁸ ≈ 3.4 × 10³⁸ possible values — a number with 39 digits. They are written as eight groups of four hexadecimal digits separated by colons, like 2001:0db8:85a3:0000:0000:8a2e:0370:7334.

To feel the gap: 4.3 billion is fewer addresses than there are people on Earth, let alone devices. 3.4 × 10³⁸ is enough to assign an address to every atom on the surface of the planet many times over. That is not a marginal upgrade; it is a different order of reality.

Why IPv6 exists: IPv4 ran out

When IPv4 was designed in the early 1980s, 4.3 billion addresses seemed inexhaustible. It was not. As the internet spread to homes, phones, and then to everything, the free pool drained. The IANA allocated the last of its central pool to the regional registries in 2011, and the Regional Internet Registries ran down their own free pools over the following decade.

The internet did not stop, because engineers had already built workarounds — chiefly NAT (Network Address Translation), which lets a whole household or even many households share a single public IPv4 address. But NAT comes at a cost: it breaks the original end-to-end design, complicates running servers, and ultimately just postpones the shortage. IPv6 was standardised to solve the problem properly, with an address space so large the question of scarcity disappears.

Notation and structure

The two formats look very different on the page.

An IPv4 address is four octets (8-bit numbers, 0–255) in decimal: 192.0.2.1. A trailing slash and number — 192.0.2.0/24 — describes a prefix, a block of addresses.

An IPv6 address is eight 16-bit groups in hexadecimal. Because they are long and often full of zeros, IPv6 allows two shorthands: leading zeros in a group can be dropped, and one run of all-zero groups can be replaced by a double colon ::. So 2001:0db8:85a3:0000:0000:8a2e:0370:7334 can be written 2001:db8:85a3::8a2e:370:7334. IPv6 networks are almost always sliced into /64 subnets, and providers typically delegate a /56 or /48 to each customer — meaning a single home can receive vastly more address space than the entire IPv4 internet contains.

Beyond size: the other differences

IPv6 is not just “IPv4 with more digits.” A few design changes matter:

• No NAT by default. With addresses no longer scarce, every device can have its own globally unique address, restoring true end-to-end connectivity. Firewalls still decide what is allowed; they simply no longer need NAT to stretch a shortage.
• A simpler, fixed header. IPv6’s main header is a fixed 40 bytes with a cleaner layout, which streamlines router processing compared with IPv4’s variable 20-byte-plus header.
• Stateless autoconfiguration (SLAAC). IPv6 devices can configure their own addresses from information the local router advertises, often without a DHCP server.
• No broadcast. IPv6 replaces IPv4’s broadcast with more efficient multicast, reducing needless network noise.

Coexistence, not replacement

There is no flag day when IPv4 switches off. The two versions coexist, and the dominant strategy is dual-stack: a device, network, or server runs both at once and uses whichever the other end supports. Where only one version is available, translation and tunneling mechanisms (such as NAT64) bridge the gap, and IPv4 itself is stretched further by carrier-grade NAT.

Adoption is real but uneven. Well over a third of traffic to major providers now arrives over IPv6, but the share varies enormously by country, by network, and by device — some mobile networks are essentially IPv6-first, while some regions remain heavily IPv4. You can see this play out in the registry data: how much IPv4 space and how many IPv6 prefixes a country has been allocated. Egypt, for example, is profiled here with its own real allocation totals:

The live card at the end of this article shows that country’s real IPv4 and IPv6 footprint from the ipdex index, with a link to its full profile and history. If an entity is not in the index, the lookup reports an honest “not found” rather than guessing.

Myth-busting

Myth: “IPv6 is just IPv4 with a bigger number.” The address space is the headline, but IPv6 also removes the need for NAT, simplifies the header, and changes how devices configure themselves. It is a redesign, not a renumbering.

Myth: “IPv6 makes the internet faster.” Not by itself. Any speed difference is situational. Your bandwidth, distance to the server, and routing dominate; the protocol version rarely does.

Myth: “We had to switch to IPv6 years ago or the internet would break.” The internet kept working because NAT and address reuse bought time. IPv6 is the durable fix, but the transition is deliberately gradual, which is why both versions are still everywhere.

Myth: “IPv6 means no more privacy because every device has a permanent address.” IPv6 includes privacy extensions that rotate a device’s address regularly. And in practice, tracking relies far more on cookies, logins, and fingerprinting than on the IP version.

Key takeaways

• IPv4 (32-bit, ~4.3 billion addresses) is the original; IPv6 (128-bit, ~3.4 × 10³⁸) was created because IPv4 ran out.
• The free IPv4 pool was exhausted — IANA in 2011, the RIRs over the following decade — but reuse and NAT keep IPv4 working.
• IPv6 also drops NAT, simplifies the header, and adds stateless autoconfiguration — it is a redesign, not just more digits.
• The two coexist through dual-stack and translation; adoption is real but varies widely by country and network.
• Registry data shows each country’s real IPv4 and IPv6 allocation — exactly what ipdex indexes.