Choosing between two proxy providers?
The same requests go through proxy A and proxy B. The report compares success rate, speed and the cost of 1,000 successful requests.
Sign in with Telegram. Free: 5 tests per 7 days, up to 60 proxies per provider.
A vs B60 lines and 1,200 requests each
Successful requestsA 94%B 81%
Response time on shared sitesA 1.4 sB 0.9 s
Working linesA 60/60B 54/60
Exit in the ordered countryA 99%B 88%
Hosting IPs in the poolA 1%B 12%
UDP via proxyA 60/60B 0/54
Cost of 1,000 successful requestsA $3.10B $3.80
Verdict. For scraping A is better: more successful requests and a lower cost per 1,000 successes. B answers faster but gets blocked more often.
What the report shows
How many requests get throughOn real sites for your task: marketplaces, social networks, search engines. With the cause of every failure.
How fastMedian and the slowest 10%, split into proxy connection, site response and download.
Where the exit really isCountry, city and network of the exit IP. We flag exits in obvious hosting and data-centre networks (by network number and name).
How the connection looks to a siteWhether the connection's system matches the browser profile and whether UDP works. Key for antidetect.
What a result costsPrice per 1,000 successful requests, not per gigabyte, so a cheap but weak proxy doesn't look like a bargain.
Which proxies don't workThe number of every dead line and the reason, ready to send to the provider for replacement.
How it works
- Paste the proxiesLines from provider A and provider B in any common format. They're checked right away.
- Pick the taskScraping, a check before antidetect, or a general check. Sites are chosen to match.
- Get the reportIn a few minutes: a one-sentence verdict, every number explained, and a PDF.
When it helps
Before buying a large volumeTest on your own task instead of trusting the provider's website.
When your current provider got worseCompare it with an alternative on the same sites at the same time.
After buying proxiesFind out whether they work and exit in the right country.
To check just your own proxies“Check one” mode: a rating for every metric, no comparison needed.
To show the result to a team or clientA PDF with numbers and a link to the method instead of “I feel like”.
One test, the same for everyone
Results can't be bought. Whoever's proxies are tested, the conditions are the same.
- Equal conditionsA and B get the same sites, at the same time, from the same server, with the same settings.
- Open methodEvery step and formula is public, versioned, with a change history. Anyone can repeat the test.
- No stretchingA difference counts as a win only if it's statistically significant. If a comparison is unreliable, the report says so.
- Results aren't for saleWe take no money for a result, a place in a report or changing one. Reports aren't edited by hand.
- Ads are separate and labelledThe Articles section may carry paid placements. They're labelled “Sponsored”, live only there and don't affect tests, reports or answers.
- Proxy credentials aren't storedThey're used only during the test and never written to the database or logs.
- Limits are statedWhat the test doesn't measure is written in the method, not hidden.
Caiman Check tests proxies as an HTTP client with a real browser fingerprint. It doesn't replace work in your own browser, but it shows the difference between providers under equal conditions.