IPv4 Address Analyzer & Class Reference for Infrastructure Engineers

Classify an IPv4 Address

Paste or type any IPv4 address. Results compute locally and announce via screen reader. Press Enter to analyze, Esc to clear.

IPv4 address

Analyze Clear

Examples

10.1.1.1 8.8.8.8 172.16.4.5 192.168.1.10 100.64.1.1 169.254.0.1 224.0.0.5 127.0.0.1 255.255.255.255

Subnet calculator

Type any IPv4 address with a CIDR prefix into the analyzer above (e.g. 192.168.1.42/24). The calculator activates the moment your input contains a /; without one, the analyzer runs as it always did. Edge cases honored: /31 point-to-point per RFC 3021 (two usable hosts, no broadcast), /32 host route (one address, zero usable), /0 default route (the full IPv4 space).

CIDR reference table

Every prefix from /8 through /32 with its subnet mask, wildcard mask, total addresses, usable host count, and a real-world example. The row matching your current calculator input is highlighted automatically.

IPv4 CIDR prefix reference. Every prefix from /8 through /32 with its subnet mask, wildcard mask, total addresses, usable host count, and a real-world example. The row matching the current calculator input is announced as the current row.

Prefix	Subnet mask	Wildcard mask	Total addresses	Usable hosts	Real-world example
/8	255.0.0.0	0.255.255.255	16,777,216	16,777,214	Class A allocation; legacy /8
/9	255.128.0.0	0.127.255.255	8,388,608	8,388,606	Half of a /8
/10	255.192.0.0	0.63.255.255	4,194,304	4,194,302	CGNAT (100.64.0.0/10)
/11	255.224.0.0	0.31.255.255	2,097,152	2,097,150	Large carrier aggregation
/12	255.240.0.0	0.15.255.255	1,048,576	1,048,574	RFC 1918 172.16.0.0/12
/13	255.248.0.0	0.7.255.255	524,288	524,286	Mid-tier ISP block
/14	255.252.0.0	0.3.255.255	262,144	262,142	Regional ISP allocation
/15	255.254.0.0	0.1.255.255	131,072	131,070	Mid-large enterprise
/16	255.255.0.0	0.0.255.255	65,536	65,534	Class B; campus net (192.168.0.0/16)
/17	255.255.128.0	0.0.127.255	32,768	32,766	Subdivided /16
/18	255.255.192.0	0.0.63.255	16,384	16,382	Large data center pod
/19	255.255.224.0	0.0.31.255	8,192	8,190	Office building / VLAN parent
/20	255.255.240.0	0.0.15.255	4,096	4,094	Office / campus VLAN parent
/21	255.255.248.0	0.0.7.255	2,048	2,046	VLAN parent / branch office
/22	255.255.252.0	0.0.3.255	1,024	1,022	~1k host VLAN
/23	255.255.254.0	0.0.1.255	512	510	~512 host VLAN; double-/24
/24	255.255.255.0	0.0.0.255	256	254	Typical LAN (Class C; default subnet)
/25	255.255.255.128	0.0.0.127	128	126	Half LAN (~126 hosts)
/26	255.255.255.192	0.0.0.63	64	62	Quarter LAN (~62 hosts)
/27	255.255.255.224	0.0.0.31	32	30	Small VLAN (~30 hosts)
/28	255.255.255.240	0.0.0.15	16	14	Mini VLAN (~14 hosts)
/29	255.255.255.248	0.0.0.7	8	6	Server cluster (~6 hosts)
/30	255.255.255.252	0.0.0.3	4	2	Point-to-point WAN (~2 hosts)
/31	255.255.255.254	0.0.0.1	2	2†	P2P link, RFC 3021 (2 hosts)
/32	255.255.255.255	0.0.0.0	1	0‡	Host route (single address)

^† RFC 3021: on a /31 point-to-point link both addresses are usable; there is no broadcast.   ^‡ A /32 identifies a single host route — no usable host count because there is no network/broadcast pair to subtract.

What an IPv4 Address Actually Is

Every IPv4 address you've ever seen — 192.168.1.10, 8.8.8.8, your home router's 192.168.0.1 — is the same thing underneath: a single 32-bit number, written for humans as four decimal numbers separated by dots. Each dot-separated number is called an octet because it's 8 bits. Four octets × 8 bits = 32 bits total. About 4.3 billion possible addresses, which sounded like infinity in 1981 and now feels claustrophobic.

Diagram 1 / 6

The IPv4 address 192.168.1.10 broken into its four octets, each shown in decimal, binary, and hexadecimal. A bracket below shows the total 32-bit length.

↑ Try any address in the analyzer above — the result panel shows the same three notations for whatever you type.

Historical IPv4 Address Classes

The IETF originally divided the IPv4 space into five classes (A–E) by the first octet. Although CIDR replaced this model for routing decisions in 1993, the classful boundaries still appear as default subnet masks, in legacy device configurations, and in the muscle memory of every network engineer. Each card below corresponds to a row in the engine above — analyze any address and the matching class is highlighted with a blue border.

The First-Octet Fingerprint

Classes A–E aren't arbitrary — they're defined by the leading bits of the first octet. The original RFC 791 designers used a clever trick: the first few bits of an address encode its class, so a router can determine the network/host split just by looking at the first 1–4 bits. Modern CIDR ignores this entirely, but the fingerprint is why 10.x.x.x is Class A and 192.x.x.x is Class C.

Diagram 2 / 6

Five rows showing the 8-bit first-octet pattern for Class A, B, C, D, and E. Each row highlights the fixed leading bits in blue and the variable bits in gray.

↑ Try 8.8.8.8 — the binary breakdown row shows 00001000, leading 0 = Class A. Try 192.168.1.1 — leading 110 = Class C.

Why Classful Addressing Existed

When IPv4 was standardized in RFC 791 (1981), the internet was a research network of a few hundred hosts. The address space was carved into rigid classes so a single byte (the first octet) determined the prefix length. A Class A allocation gave one organization 16.7 million addresses. A Class B gave 65,534. A Class C gave 254. There was no in-between.

That model worked for a few years. By the late 1980s, allocations were being granted to organizations that needed maybe 3,000 hosts — far more than a Class C (254) but a tiny fraction of a Class B (65k). The IETF watched the IANA pool drain faster than usage actually demanded. The fix arrived as CIDR in 1993.

RFC 1918 and the Private Address Economy

RFC 1918 (1996) carved three ranges out of the global address space for unrestricted internal use: 10.0.0.0/8, 172.16.0.0/12, and 192.168.0.0/16. These addresses are never advertised on the public internet. In exchange, every organization can reuse them inside their own network without coordination. The bargain that makes this work is NAT — the gateway box that translates internal private addresses into a public one when traffic crosses to the internet.

Diagram 3 / 6

Three internal devices with private 192.168.1.x addresses on the left connect through a NAT router in the center, which translates their traffic to a single public IP on the right.

↑ Try 192.168.1.47 — the analyzer flags it as Private RFC 1918 (192.168/16), Not internet-routable. Then try 73.84.5.12 — flagged as Public, Internet-routable.

CGNAT — The Hidden Fourth Private Range

By 2010, the original RFC 1918 ranges weren't enough for the largest ISPs. The IETF responded with RFC 6598: a fourth private-style range, 100.64.0.0/10, reserved specifically for Carrier-Grade NAT. With CGNAT, multiple subscribers sit behind a single ISP-owned NAT, sharing one public IP between many homes.

Diagram 4 / 6

Stacked horizontal bands show the path from a customer's device through their home NAT, then the ISP's CGNAT, to the public internet.

Special Ranges — Why They Matter at 3 a.m.

APIPA (169.254.0.0/16): A host self-assigns from this range when DHCP fails. If your monitoring sees a server with a 169.254 address, you have a DHCP outage, a missing relay, a VLAN misconfig, or a switchport in the wrong access VLAN.

Loopback (127.0.0.0/8): Internal to the host. Useful for binding daemons that should never accept external connections. If your load balancer points at 127.0.0.1, your service is unreachable from elsewhere — either deliberate or a misconfig.

Multicast (224.0.0.0/4): Group communication. Routing protocols (OSPF at 224.0.0.5/6, EIGRP at 224.0.0.10), service discovery (mDNS at 224.0.0.251), replication traffic. Routers do not forward multicast by default — PIM, IGMP, and MSDP are required.

TEST-NET (192.0.2.0/24, 198.51.100.0/24, 203.0.113.0/24): Documentation-only ranges per RFC 5737. If you ever see one of these in production, someone copy-pasted a tutorial without changing the example values.

Benchmarking (198.18.0.0/15): Reserved per RFC 2544 for testing routing and switching hardware throughput and latency. Lab-only.

Broadcast (255.255.255.255): Limited-scope broadcast. Used by DHCP DISCOVER, BOOTP, and a handful of legacy auto-discovery protocols. Routers drop it by default.

Unspecified (0.0.0.0): Context-dependent. As a destination, it means "default route" (or "any" in some socket APIs). As a source, it usually means "I don't have an address yet" (DHCP DISCOVER source).

↑ Try each one in the analyzer — the operational-context callout in the result panel explains when you'll see this address in the wild.

CIDR — The Model That Actually Runs the Internet

CIDR (RFC 4632) replaced classful routing in 1993. The shift was conceptual: instead of the first octet implying the prefix length, the prefix length is explicit (/24, /16, /22) and can be any value from 0 to 32. This enabled route aggregation — an ISP can advertise 203.0.113.0/22 as a single BGP entry covering four contiguous /24s, instead of four separate routes.

Diagram 5 / 6

Left side shows three large blocks representing the only available prefix sizes under classful. Right side shows many flexible blocks of arbitrary size representing CIDR.

Packet routing — NAT vs CGNAT

A packet from your laptop to a public endpoint traverses one NAT in a typical home or small office. When your ISP places you behind Carrier-Grade NAT, that becomes two NATs in series — and inbound port forwarding stops being possible from the customer side.

Diagram 6 / 6

For the engineer

CGNAT was the IPv4 stopgap when RIRs ran out of free /8 allocations around 2011–2015. The carrier reserves a /10 from RFC 6598 (100.64.0.0/10) for the inside of the carrier NAT. If you are diagnosing a customer who insists their port forward "isn't working anymore" and they recently switched ISPs or moved to fiber, run curl https://ifconfig.me on a host inside their network and check whether the result is in 100.64.0.0/10 vs their public provider block. If it is CGNAT, the only path to inbound is a VPN or tunneling provider — the customer has no way to punch through the carrier NAT.

This will bite you

Many ISPs do not disclose CGNAT in their marketing or onboarding. Customer-facing speed tests work fine, web browsing works fine, Netflix works fine. The breakage is asymmetric and only affects services the customer hosts. Confirm CGNAT status by comparing "what is my IP" output between an internal host and the ISP's customer portal — if they differ AND the host's IP is in 100.64.0.0/10, the customer is behind CGNAT.

The internet's routing tables are physically possible today because of CIDR. Under classful routing, the global table would have exhausted router memory years ago. For day-to-day engineering, CIDR means: think in prefix lengths, not in classes. A /27 is a /27 whether it sits in former Class A, B, or C space.