Proxy IP checker▮

This is the most important column. Every IP belongs to an ASN, a public registry record naming the network that owns it, and any website can read it instantly. Datacenter means the ASN belongs to a hosting company such as AWS, OVH, Hetzner or M247. Cheap and fast, but anyone who looks can see they are not residential. Residential-looking means the ASN belongs to a consumer internet provider, so the address resembles an ordinary home connection. Mobile means a cellular carrier, which is the hardest category for a platform to act against, because carriers put large numbers of real subscribers behind small shared address pools.

One honest caveat about confidence: datacenter identification is close to certain, because hosting ASNs are unambiguous and self-declared in registry data. Residential identification is an inference from the owning organisation looking like a consumer ISP. A hosting reseller operating inside an ISP's address space can read as residential here, which is exactly why it is sold as a residential proxy.

A /24 is a block of 256 consecutive addresses, and it is normally allocated to a single customer of a single provider. Addresses inside one are therefore closely related, and risk systems routinely treat a /24 as one entity, not 256 independent ones. Buying fifty proxies that all live in 45.155.204.x gives you fifty addresses but, in practice, close to one identity. This page counts distinct /24s alongside distinct IPs so the gap is visible: if you have 50 proxies across 2 subnets, the second number is the one that matters.

A list containing 1.2.3.4:8000, 1.2.3.4:8001 and 1.2.3.4:8002 looks like three proxies and is one address. Ports are invisible to the destination. A website sees the IP the traffic arrives from and nothing else. Sellers sometimes pad list sizes this way, so the tool flags reused IPs explicitly. They are still useful for separating sessions on your side, but they provide no separation at all on the side that matters.

Add the advertised country to the end of any line after a vertical bar, like 84.247.60.125:6095 | Poland, and the tool compares it against the registry record for that address. Two-letter codes and full names both work. Everyday names are handled too, so UK is understood as GB and USA as US; without that, every British proxy would be reported as wrong purely because the ISO code differs from the name people actually use.

A mismatch means the address is not registered where it was sold to you, which matters if you are buying for geo-targeting. Bear in mind that geolocation is inference, not measurement, so a single-city difference inside one country is weak evidence and a different country is strong evidence. Only country-level differences are flagged here for that reason.

Each IP carries a timezone from its geolocation record, and the tool compares it against the timezone your own browser reports. A gap is not automatically a problem. It is expected if you are testing a foreign proxy from home. It matters when the device that will actually use the proxy has a clock set to a different region than the exit IP, because that inconsistency is directly readable and is a well-established signal. If you are running these proxies on phones, the phones' timezone is the one to compare, not this browser's.

Entries flagged private or bogon use address ranges that cannot exist on the public internet: 10.x, 192.168.x, 172.16–31.x, 127.x and 169.254.x. They are only reachable inside a local network, so as an entry in a purchased proxy list they indicate either a copy-paste error or a seller padding the file. Entries flagged hostname use a domain instead of an IP; those cannot be looked up here, because resolving a hostname to an address needs a DNS query that a browser will not perform for an arbitrary name.

No, and it is worth understanding why, because it also tells you what to trust elsewhere. Checking liveness means opening a TCP connection and speaking HTTP or SOCKS. A SOCKS5 handshake starts by writing the raw bytes 05 01 00 to a socket, and browsers provide no way to write raw bytes to a socket at all. fetch() deliberately has no proxy option either. So a browser-only page cannot do it, and any tool reporting latency is doing the work on its servers. That is a fine architecture, but it means the latency number is measured from that company's datacenter, not from your phones. A proxy that looks fast there can be slow where you actually use it.

Because working and trusted are different properties. A proxy can pass every liveness test. It connects, it is fast, it hides your address. And it can still be pre-judged before you send a single request, purely on who owns it. If the ASN belongs to a hosting provider, the address begins with an elevated risk score, since almost no ordinary person browses from a rented server. Reputation history compounds it: addresses are recycled, so whoever held it before you leaves a trace. This is why an ASN column belongs on a proxy checker and why liveness alone tells you little about whether an address will be trusted.

Not for this use case, and the nuance matters. The most widely used blocklists are built for email spam filtering. One of them, the Policy Block List, exists specifically to mark address ranges that providers have designated as dynamic residential, the sort of address a home router receives. A generic blacklist tool reports that as a negative. For someone deliberately seeking residential-looking addresses it is close to confirmation that the range is what it claims to be. Read any blacklist result in the context of what you are trying to do with the address, not as a single verdict.

The list itself is parsed instantly and there is no limit on that. Formatting, duplicates, subnet grouping and private-range detection all happen locally and cost nothing. The network lookups are the constrained part, since they use a free public API with a monthly quota. The tool therefore looks up each unique IP once instead of each line, caches results in your browser for seven days so repeat checks are free, and stops at 256 unique addresses per run. For larger lists, split them or re-run: anything already cached will not be requested again.

They are parsed so the tool can understand the line format, displayed nowhere, stored nowhere, and never included in any outbound request. Only the bare IP address is sent to the lookup API, because that is all a network lookup needs. Nothing is written to disk except the IP-to-network cache, which contains no credentials. Open your developer tools and watch the Network tab during a run if you would rather verify it than take our word for it.