Paste up to 500 proxies in whatever format your provider handed you. The parser normalises the lot, flags the lines that can never work, and tests the rest concurrently for liveness, latency, exit country, ASN, anonymity and fraud score.
Used only for lines that carry no scheme of their own.
A checker is only worth the confidence you can put in its verdicts. Here is precisely what each one is measuring, and where it stops.
A proxy is alive here if a full request completes through it to the chosen target and returns a usable response. That is a stricter test than a TCP connect, which is what most checkers measure — a socket that opens and then hangs on the first byte will pass a connect test and fail every real job. Latency is measured from connect to first byte, so it includes the exit's own handshake, not just the round trip to it.
Addresses in private, loopback, link-local, multicast and reserved ranges cannot be reached from the public internet, so testing them wastes a slot and tells you nothing. They are marked skipped with the reason attached. Carrier-grade NAT space (100.64.0.0/10) is flagged but still tested — it is unusual in a proxy list and usually indicates a misconfigured export.
Fifty IPs on three /24s is not fifty proxies for anything that rate-limits by subnet, which is most of what matters. The summary reports unique /24s alongside the raw count, because that ratio is the difference between a list that survives a release and one that gets range-banned in the first minute.
The score estimates how likely a target is to treat the exit as suspicious, based on its reputation history, its classification as datacentre or residential, and its presence on public blocklists. It is a prediction, not a verdict: a clean score does not guarantee access, and a poor one does not guarantee a block. Read it as triage on a large list, not as a decision on a single IP.
Paste the list as it came. The parser detects the shape line by line, so a file that mixes three
formats — as exports routinely do — still comes through whole. Blank lines and lines starting # or // are ignored.
| Shape | Example | Notes |
|---|---|---|
| host:port | 154.12.88.31:8080 | Open proxy, no auth |
| host:port:user:pass | 154.12.88.31:8080:jchen:h4nd0ff | The most common paid-provider export |
| user:pass@host:port | jchen:[email protected]:8080 | cURL / env-var style |
| scheme://user:pass@host:port | socks5://jchen:[email protected]:1080 | Fully qualified; scheme wins over the default |
| host port user pass | 154.12.88.31 8080 jchen h4nd0ff | Space, tab, comma, semicolon and pipe all work |
| [v6]:port | [2a02:c7f:e01d::1]:3128 | IPv6 must be bracketed |
The same engine converts on the way out. Check a list in one format, export the survivors in another — that is the whole of what a standalone format converter does, and it is built in here.
The level is determined by which headers survive the hop. It is the single most useful column in the table, and the one most often misread.
500 per run. That is a browser limit rather than a licensing one — beyond roughly that many concurrent checks the tab, not the network, becomes the bottleneck. There is no daily cap, no account and no queue.
No. Parsing, validation and formatting happen entirely in your browser. When you press Check, each proxy is sent to the checking service for exactly as long as the test takes and is never written to disk. Credentials are masked in the results table by default so the output is safe to screenshot.
HTTP, HTTPS, SOCKS4, SOCKS5 and SOCKS5h. If a line carries no scheme the tool applies the default you pick above the input, and says so in the notes for that row.
Three usual causes. It is IP-authorised rather than user/password authorised, and our checker's address is not on your allowlist. It is alive but blocks the specific target you chose — try a different one. Or it is slower than your timeout, which the default of 10 seconds will catch on genuinely marginal exits.
Elite sends no proxy headers, so the origin cannot tell a proxy is in the path. Anonymous announces itself as a proxy but hides your address. Transparent forwards your real address in X-Forwarded-For and offers no anonymity at all.
Yes. Write them as socks5://user:pass@host:port, or paste host:port:user:pass and set the default scheme to SOCKS5.
Under 250 ms is fast enough that the proxy is not your bottleneck. 250–600 ms is normal for a residential exit and fine for scraping. Above 600 ms starts to hurt anything interactive, and above two seconds the exit is usually oversubscribed rather than distant.
Not necessarily. This tool tests reachability and reports the exit's properties. Whether a specific site admits it depends on that site's anti-bot stack and the reputation of the subnet, which is a different question from whether the proxy is alive.
Dedicated ISP proxies on named carrier ASNs do not churn between runs, and the subnet spread is published before you buy. Test ours with this same tool — the trial is $1.
Checking one address rather than a list? The fraud score checker reports reputation, blocklist status and classification for a single IP.