IPv6 Address Types Explained: Unicast, Multicast, and Anycast (2026 Guide)

Last updated: July 2026 · Reviewed against: RFC 4291, RFC 4193, and official Cisco IOS documentation.

IPv6 replaces IPv4’s aging 32-bit addressing with a 128-bit scheme that provides roughly 340 undecillion addresses. But the address space is only half the story. To design, configure, and troubleshoot modern networks, you need to understand how IPv6 addresses are classified and what each type is used for.

This guide breaks down the three IPv6 address categories — unicast, multicast, and anycast — along with the special-purpose addresses every network professional and CCNA candidate should know. Every prefix and notation here has been verified against the relevant RFCs.

---

Quick Comparison Table

Feature	Unicast	Multicast	Anycast
Delivery model	One-to-one	One-to-many	One-to-nearest
Recipients	A single interface	All interfaces in a group	The closest of several interfaces
Identifying prefix	2000::/3 (GUA), FE80::/10 (link-local), FC00::/7 (ULA)	FF00::/8	Uses unicast address space
Replaces broadcast?	No	Yes (IPv6 has no broadcast)	No
Typical use	Standard host-to-host traffic	Routing protocols, service discovery	DNS root servers, CDNs, load distribution

Note: IPv6 has no broadcast address. Any function that relied on broadcast in IPv4 is handled by multicast in IPv6.

The Three IPv6 Address Categories

Unlike IPv4, which defines unicast, multicast, and broadcast, IPv6 defines exactly three types:

• Unicast — identifies a single interface. A packet sent to a unicast address is delivered to that one interface. This is the everyday, host-to-host address type.
• Multicast — identifies a group of interfaces, typically on different nodes. A packet sent to a multicast address is delivered to every interface in the group.
• Anycast — identifies a set of interfaces (usually on different nodes) that share the same address. A packet sent to an anycast address is delivered to the nearest interface in the set, as determined by the routing protocol’s distance metric.

The sections below examine each in detail.

Unicast Address

A unicast address uniquely identifies a single interface on an IPv6-enabled device. There are several kinds of unicast address, each with a defined scope.

The types of IPv6 unicast are:

Global unicast addresses (GUA), link-local addresses, and unique local IPv6 unicast addresses. The most common are global unicast and link-local unicast addresses. The figure below illustrates IPv6 Unicast address types:-

Global Unicast Addresses (GUA)

A Global Unicast Address is the IPv6 equivalent of a public IPv4 address: globally unique and routable across the internet.

Global unicast addresses currently come from the 2000::/3 range — that is, any address whose first three bits are 001, spanning 2000:: through 3FFF:FFFF:.... A common misconception is that only 2001::/16 is allocated. In reality, the Regional Internet Registries have assigned many blocks within 2000::/3, including 2400::/12 (APNIC), 2600::/12 (ARIN), 2800::/12 (LACNIC), 2a00::/12 (RIPE NCC), and others.

Structure of a global unicast address (128 bits total):

Portion	Length	Purpose
Global Routing Prefix	48 bits (typical)	Assigned to an organization by an ISP/RIR; identifies the site
Subnet ID	16 bits (typical)	Chosen by the organization to identify a subnet within the site
Interface ID	64 bits	Identifies the specific interface on the subnet

A typical site therefore receives a /48 prefix, uses the next 16 bits for internal subnetting (allowing 65,536 subnets), and reserves the final 64 bits for the interface ID. The /64 interface portion is the standard for almost all IPv6 subnets and is required for SLAAC (Stateless Address Autoconfiguration) to work.

A typical site therefore receives a /48 prefix, uses the next 16 bits for internal subnetting (allowing 65,536 subnets), and reserves the final 64 bits for the interface ID. The /64 interface portion is the standard for almost all IPv6 subnets and is required for SLAAC (Stateless Address Autoconfiguration) to work.

Common point of confusion: The subnet ID and the interface ID are different fields. The subnet ID (16 bits in the typical layout) selects a subnet; the interface ID (64 bits) identifies the host within it. Do not conflate the two.

Example: 2001:0db8:acad:0001:0000:0000:0000:0100

• Global routing prefix: 2001:0db8:acad
• Subnet ID: 0001
• Interface ID: 0000:0000:0000:0100

IPv6 Link-local Addresses

A link-local address is used for communication confined to a single link (network segment). These addresses are never routed beyond the local link.

• Prefix: FE80::/10 — the first 10 bits are fixed. In practice the bits between the /10 boundary and the 64-bit interface ID are set to zero, so link-local addresses effectively begin FE80::/64.
• Every IPv6-enabled interface automatically generates a link-local address, whether or not any other address is configured.
• They are essential for neighbor discovery, router solicitation, and as next-hop addresses in routing protocols.

Best practice: Because interfaces generate link-local addresses automatically, you rarely need to configure them manually. On routers, however, administrators often assign a short, memorable link-local address (for example FE80::1) to simplify routing configuration and troubleshooting.

Example: FE80::BE85:56FF:FE60:ED75

Tip: Never manually assign link-local addresses—devices generate them automatically.

IPv6 Unique Local Addresses

Unique Local Addresses are the IPv6 counterpart to IPv4 private addresses (like 10.0.0.0/8 or 192.168.0.0/16). They are routable within an organization but are not routable on the public internet.

• Prefix: FC00::/7. This single /7 block spans every address from FC00:: through FDFF:FFFF:....
• In practice, the FD00::/8 half is used for locally assigned addresses, where the organization generates a random 40-bit Global ID to make collisions between sites extremely unlikely. The FC00::/8 half is reserved and not currently in general use.

Correction on a common error: The ULA range is a single prefix, FC00::/7. It is incorrect to describe it as “FC00::/7 to FDFF::/7” — that notation treats two addresses as if each were its own /7 block, which is meaningless. FC00::/7 already includes everything up to FDFF:FFFF:....

Example: FD00:1234:5678:9abc::1

Historical note: Previous version of the article may reference site-local addresses (FEC0::/10). These were deprecated by RFC 3879 and replaced by Unique Local Addresses. Do not use site-local addresses in new designs.

---

Special Unicast Addresses

Two special unicast addresses appear constantly in practice:

• Unspecified address — :: (all 128 bits zero, written ::/128 when expressed as a prefix). It indicates the absence of an address. A host uses :: as the source address while it is still acquiring a real address (for example, in a Duplicate Address Detection message). It must never be assigned to an interface as a usable address. Correction on a common error: The unspecified address is ::, not ::/0. The prefix ::/0 is the IPv6 default route — an entirely different concept. Confusing the two is a frequent exam trap.
• Loopback address — ::1 (written ::1/128). This is the IPv6 equivalent of IPv4’s 127.0.0.1. A device uses it to send packets to itself for testing the local TCP/IP stack. It is never assigned to a physical interface and never appears as a source or destination on the wire.

Embedded IPv4 Addresses

During the long transition from IPv4 to IPv6, several mechanisms embed an IPv4 address inside an IPv6 address:

• IPv4-mapped IPv6 addresses — ::ffff:a.b.c.d (for example ::ffff:192.0.2.128). These represent an IPv4 node to an IPv6 application. This is the form you will encounter in modern dual-stack systems.

Important correction: Older tutorials often present IPv4-compatible addresses in the form ::a.b.c.d (for example ::192.0.2.128) as if they were current. They are not. IPv4-compatible addresses were deprecated by RFC 4291 and should not be used. Modern transition relies on IPv4-mapped addresses plus translation and tunneling technologies such as NAT64, 6to4, and Teredo.

---

Multicast Addresses

A multicast address identifies a group of interfaces. A packet sent to a multicast address is delivered to every member of that group. Multicast is central to IPv6 — it takes over every role that broadcast played in IPv4.

• Prefix: FF00::/8 — any address beginning with FF is multicast.

The second byte of a multicast address encodes flags and scope, which determine how far the multicast traffic may travel (interface-local, link-local, site-local, organization-local, or global).

Well-known multicast addresses you should memorize:

Address	Group
FF02::1	All nodes on the link
FF02::2	All routers on the link
FF02::5	All OSPFv3 routers
FF02::6	All OSPFv3 designated routers
FF02::9	All RIPng routers
FF02::A	All EIGRP (IPv6) routers
FF02::1:FFxx:xxxx	Solicited-node multicast (used by Neighbor Discovery)

The solicited-node multicast address deserves special attention: it is automatically created for every unicast and anycast address on an interface and is used by the Neighbor Discovery Protocol to resolve link-layer addresses — the job ARP performed in IPv4.

---

Anycast Addresses

An anycast address is assigned to multiple interfaces, typically on different devices. When a packet is sent to an anycast address, the network routes it to the nearest interface holding that address, where “nearest” is defined by the routing protocol’s metric.

• Anycast addresses are allocated from the normal unicast address space — there is no distinct anycast prefix. An address becomes anycast simply because it is configured on more than one interface and the devices are told to treat it as anycast.
• Because several interfaces share the address, anycast is not used as a packet’s source address.

Common uses:

• DNS root and recursive servers — the same IP is announced from many locations worldwide, so users reach the closest instance.
• Content Delivery Networks (CDNs) — direct users to the nearest edge server, reducing latency.
• Load distribution and redundancy — if the nearest node fails, routing automatically shifts traffic to the next-closest node.

---

Building an Interface ID with EUI-64

The Modified EUI-64 process derives a 64-bit interface ID from a 48-bit MAC address. It is worth understanding because SLAAC can use it to generate addresses automatically.

Worked example — convert the MAC address BC:85:56:60:ED:75 into an interface ID:

1. Split the MAC in half and insert FF:FE between the two halves: BC:85:56 + FF:FE + 60:ED:75 → BC:85:56:FF:FE:60:ED:75
2. Flip the 7th bit of the first byte (the Universal/Local bit). The first byte is BC = 1011 1100 in binary. Inverting the 7th bit (counting from the left) changes 1011 1100 to 1011 1110 = BE.
3. Write the result in IPv6 notation: BE85:56FF:FE60:ED75

Combined with a /64 prefix such as FE80::/64, the full link-local address becomes:

FE80::BE85:56FF:FE60:ED75

Privacy note: Because EUI-64 embeds the hardware MAC address, it can expose a device’s identity across networks. Most modern operating systems therefore default to privacy extensions (RFC 4941) or randomized interface identifiers instead of EUI-64.

---

How to Identify an IPv6 Address Type

Use this quick decision guide based on the leading bits or hextets:

• Starts with FF → Multicast (FF00::/8)
• Starts with FE80 → Link-local unicast (FE80::/10)
• Starts with FC or FD → Unique Local unicast (FC00::/7)
• Starts with 2 or 3 (first hextet 2000–3FFF) → Global unicast (2000::/3)
• Exactly ::1 → Loopback
• Exactly :: → Unspecified
• Contains ::ffff: followed by dotted-decimal → IPv4-mapped
• Shares a unicast address across multiple nodes → Anycast (no unique prefix)

---

Verifying Addresses on Real Devices

Reading about address types is one thing; seeing them on live equipment cements the concepts.

On Windows (ipconfig):

> ipconfig
   IPv6 Address. . . . . . . . . . . : 2001:db8:acad:1::100
   Link-local IPv6 Address . . . . . : fe80::be85:56ff:fe60:ed75%12

On Linux or macOS (ip -6 addr / ifconfig):

$ ip -6 addr show dev eth0
    inet6 2001:db8:acad:1::100/64 scope global
    inet6 fe80::be85:56ff:fe60:ed75/64 scope link

On a Cisco router (show ipv6 interface):

Router# show ipv6 interface GigabitEthernet0/0
GigabitEthernet0/0 is up, line protocol is up
  IPv6 is enabled, link-local address is FE80::BE85:56FF:FE60:ED75
  Global unicast address(es):
    2001:DB8:ACAD:1::1, subnet is 2001:DB8:ACAD:1::/64
  Joined group address(es):
    FF02::1
    FF02::2
    FF02::1:FF00:1

Notice how a single interface simultaneously holds a link-local address, a global unicast address, and membership in several multicast groups — including the solicited-node group (FF02::1:FF00:1) derived from its unicast address.

---

Frequently Asked Questions

What are the three types of IPv6 addresses?

IPv6 defines unicast (one-to-one), multicast (one-to-many), and anycast (one-to-nearest). Unlike IPv4, IPv6 has no broadcast address; multicast handles those functions.

Does IPv6 have a broadcast address?

No. IPv6 eliminates broadcast entirely. Every task that used broadcast in IPv4 — such as address resolution — is handled by multicast in IPv6, for example the all-nodes group FF02::1.

What is the difference between a global unicast and a unique local address?

A global unicast address (2000::/3) is globally unique and routable on the public internet, like a public IPv4 address. A unique local address (FC00::/7) is routable only within an organization, like a private IPv4 address.

What is the IPv6 loopback address?

‘It is ::1 (equivalently ::1/128), the IPv6 equivalent of IPv4’s 127.0.0.1. It lets a device send packets to itself to test the local TCP/IP stack.

Is ::/0 the unspecified address?

No. The unspecified address is ::. The prefix ::/0 is the IPv6 default route. They are commonly confused but represent entirely different things.

Are IPv4-compatible IPv6 addresses still used?

No. IPv4-compatible addresses (::a.b.c.d) were deprecated by RFC 4291. Modern systems use IPv4-mapped addresses (::ffff:a.b.c.d) together with translation and tunneling technologies such as NAT64 and 6to4.

What is a solicited-node multicast address?

It is a multicast address in the form FF02::1:FFxx:xxxx, automatically generated from an interface’s unicast or anycast address. Neighbor Discovery uses it to resolve link-layer addresses, replacing IPv4’s ARP.

Why do DNS root servers use anycast?

Anycast lets many physical servers share one IP address. Each user is routed to the nearest server, improving speed and resilience — if one node fails, traffic automatically shifts to the next-closest one.

---

Key Takeaways

• IPv6 defines three address types: unicast (one-to-one), multicast (one-to-many), and anycast (one-to-nearest). There is no broadcast.
• Global unicast addresses come from 2000::/3 and are internet-routable; a site typically gets a /48 with a 16-bit subnet ID and a 64-bit interface ID.
• Link-local addresses (FE80::/10) are automatic and confined to a single link.
• Unique local addresses use the single prefix FC00::/7 — not “FC00::/7 to FDFF::/7.”
• The unspecified address is ::; the loopback address is ::1. Neither is ::/0, which is the default route.
• Prefer IPv4-mapped addresses (::ffff:a.b.c.d); IPv4-compatible addresses are deprecated.
• Multicast (FF00::/8) replaces broadcast and powers Neighbor Discovery via solicited-node addresses.
• Anycast uses ordinary unicast space and underpins DNS root servers, CDNs, and load distribution.

---

References

• RFC 4291 — IP Version 6 Addressing Architecture
• RFC 4193 — Unique Local IPv6 Unicast Addresses
• RFC 3879 — Deprecating Site Local Addresses
• RFC 4941 — Privacy Extensions for Stateless Address Autoconfiguration
• Cisco IOS IPv6 Configuration Guide

---

Written and technically reviewed for accuracy against current IETF standards. Found an error or have a question? Leave a comment below.