IPv4 vs IPv6: Complete Technical Comparison (2026)

IPv4 vs IPv6: Complete Comparison

IPv4 Address Space: Why It Ran Out

IPv4 uses 32-bit addresses, providing 2³² = 4,294,967,296 total addresses. Subtract reserved ranges (RFC 1918 private addresses, loopback, multicast, reserved blocks) and the usable public address space shrinks to roughly 3.7 billion addresses. IANA distributed the last /8 blocks to the five RIRs in February 2011; APNIC (Asia-Pacific) exhausted its allocation first in April 2011, followed by other RIRs through 2019.

The internet continued growing despite address exhaustion through several mechanisms: NAT (Network Address Translation) allows thousands of devices to share a single public IP; CGNAT (Carrier-Grade NAT) extends this to entire ISP customer bases; IP address trading markets emerged; and aggressive reclamation of historically over-allocated legacy address blocks returned some space.

IPv6: 340 Undecillion Addresses

IPv6 uses 128-bit addresses: 2¹²⁸ = approximately 3.4 × 10³⁸ addresses. To put this in perspective: there are roughly 10⁸⁰ atoms in the observable universe. IPv6 provides roughly 4.8 × 10²⁸ addresses per human on Earth — enough to give every grain of sand on every beach its own address, multiplied many times over.

IPv6 addresses are written as eight groups of four hexadecimal digits separated by colons: 2001:0db8:85a3:0000:0000:8a2e:0370:7334. Leading zeros within groups can be omitted, and one continuous sequence of all-zero groups can be replaced with ::, yielding 2001:db8:85a3::8a2e:370:7334.

IPv6 Address Types

IPv6 Improvements Over IPv4

Fixed header size: IPv4 headers are 20–60 bytes (variable due to options). IPv6 headers are always exactly 40 bytes, eliminating variable-length parsing overhead and enabling faster router processing.

No header checksum: IPv4 includes a header checksum that must be recalculated at every hop (since TTL decrements). IPv6 removes this, improving forwarding performance and relying on layer 2 and transport-layer checksums for error detection.

SLAAC (Stateless Address Autoconfiguration): IPv6 devices can automatically configure their own globally routable addresses using their network prefix (from router advertisements) and their interface identifier — eliminating DHCP dependency for basic connectivity.

Extension headers: IPv6 uses a chain of optional extension headers rather than a variable options field in the main header. This allows new features to be added without changing the core header format.

Source-only fragmentation: In IPv4, routers can fragment packets that exceed the MTU. In IPv6, only the source host fragments — routers that can't forward a packet due to MTU issues send an ICMPv6 "Packet Too Big" message instead. This eliminates fragmentation overhead on routers.

Dual Stack: How IPv4 and IPv6 Coexist

Most modern networks and devices operate in "dual-stack" mode — simultaneously supporting both IPv4 and IPv6. When a device connects to a dual-stack network, it receives both an IPv4 and an IPv6 address. When connecting to a website, the OS uses an algorithm (Happy Eyeballs / RFC 6555) to attempt both IPv4 and IPv6 connections simultaneously, using whichever connects first.

Since IPv6 is generally preferred when available (Happy Eyeballs slightly favors it), dual-stack websites often see most of their traffic from modern clients arriving over IPv6. You can check whether your current connection uses IPv4 or IPv6 with our My IP Address tool, and look up the ASN and routing details of any IPv6 address with our IPv6 Lookup tool.

IPv6 Adoption Status (2026)

IPv6 adoption has accelerated significantly but remains uneven. Google's IPv6 statistics (one of the most reliable public metrics) show approximately 35–40% of Google users connecting via IPv6 globally as of 2026. Mobile networks have driven much of this growth — major carriers in the US, India, Germany, and Japan now deploy IPv6 by default for mobile subscribers.

Enterprise adoption lags consumer ISPs. Many corporate networks still rely on IPv4 internally, with IPv6 only deployed at the edge or not at all. Legacy equipment compatibility, IT team familiarity, and the "good enough" nature of NAT/IPv4 for internal networks continue to slow enterprise migration.

FAQs: IPv4 vs IPv6