Skip to content

How to Vet a Proxy Provider: A Hands-On Quality Check

Test a proxy provider with a small paid top-up and two free checkers: the terms to read first, what each result means, and the red flags.

HProxy TeamJuly 3, 2026Updated October 23, 202610 min read

How to Vet a Proxy Provider Before You Buy

A seller's homepage can promise anything: millions of IPs, high success rates, clean addresses. The proxies themselves can be checked. This guide walks through a test you can run before you buy more than a small amount from any seller. Read the terms, top up a little, check what you got with two free tools, and hold the results against the red flags at the end.

What do you need for the test?

  • The smallest top-up the seller sells, bought after you read its terms.
  • 10 to 20 of its proxies in the country you plan to use.
  • Our free online proxy checker, plus our free proxy checker app for proxies that only work from your own IP.
  • One page from the site you plan to use the proxies on.

The test costs you the top-up and a little of its traffic. Neither checker needs an account.

What should you read before you top up?

Four terms decide whether a small test is safe: the refund rule, how long traffic lasts, the identity checks and the forbidden uses.

The refund rule matters most, because it decides whether you can get your test money back. We hold records of the refund terms of 39 of the 57 sellers in our 10 GB price study, read from their own pages between September 26 and October 10, 2026.

What 39 sellers' refund terms allowed on unused residential traffic
No refund, or only for a faultor for a billing error
18
A window of 1 to 3 dayswith a cap on use
7
A window of 7 to 14 dayswith a cap on use
6
30 days or moreunused or lightly used traffic
4
Its own pages disagreethe terms against a FAQ or homepage
4
18 of 39 sellers allowed no refund for unused traffic, and 7 of the 17 windows closed within three days.
Source: Our records of each seller's own refund page or terms, read through our server September 26 to October 10, 2026; the 39 of the 57 sellers in our 10 GB price study whose refund terms we hold · hproxy.comProxy.

Almost every window came with a cap on use, from 0.1 GB of traffic to half the plan. Twelve of the 39 sellers refused refunds on crypto payments, or paid back only card and PayPal payments. Four said one thing in their terms and another in a FAQ, a help center or on the homepage. If you meet that, ask support which rule applies and keep the answer.

The other three terms:

  • How long traffic lasts. In our price records from late September 2026, three sellers let residential traffic or credits lapse after 30 days, one calendar month or 60 days, and five said their traffic never expires. Make sure your test traffic will still be there when you finish. Our guide to reading a proxy pricing page lists the rules we recorded, by name.
  • Identity checks. Of 15 sellers whose identity rules we hold, five asked for an identity check, a questionnaire or an application before the first purchase, and eight kept the right to ask later. One large seller opens residential access only to registered companies after a review, as our comparison with Bright Data shows. Find out before you plan a project around a seller.
  • Forbidden uses. Read which sites and uses the seller blocks, so the job you have in mind is one it allows.

Read, too, what each seller publishes about where its addresses come from.

How do you get proxies to test from a small top-up?

Buy the smallest amount the seller sells, in the country you plan to use. Then turn it into a list of proxy lines. If the dashboard can generate sticky sessions, generate 10 to 20 of them, one line each, so every line keeps its own exit for a while. Identical lines show one result in the checker, so give each line its own session. Rotating and static residential plans behave differently here, so test the type you will buy.

Write the lines in a format the checker reads, with your own login in place of the placeholders:

HOST:PORT:YOUR-LOGIN:YOUR-PASSWORD
YOUR-LOGIN:YOUR-PASSWORD@HOST:PORT

How do you check the proxies with the online checker?

Paste the lines into our online proxy checker and start the check. It takes up to 100,000 lines a run with no account, tries HTTP, HTTPS, SOCKS4 and SOCKS5 on every line, and fills one row per proxy as the answers come back. Each row reports:

  • Status: working, dead or timed out, from a real connection opened through the proxy at that moment.
  • Protocol: which of the four the proxy speaks.
  • Anonymity: elite, anonymous or transparent, read from the headers the proxy forwards. A transparent proxy forwards your real IP.
  • Location: the country and city of the exit, the address websites see. For a gateway it differs from the address you pasted.
  • Network: the internet provider and the network number (ASN) behind the exit, with a flag when that network belongs to a hosting company.
  • Speed: the round trip in milliseconds, the fastest of the protocol tests that answered.

The online checker connects from our servers, and what you paste never joins our free proxy list. If the seller only accepts requests from your own IP, every line will fail there, so use the app.

What does the checker app add?

Our free proxy checker app, for Windows, macOS, Linux and Android, runs the check from your own connection. That makes it the way to test a proxy locked to your IP. It adds two things:

  • Connect sends your computer through one of the proxies. Open the page you plan to use, the way you will use it. While you are connected, the app shows the address you appear as and checks the connection every twenty seconds.
  • The fraud score rates the address you come out of, from 0 to 100, through a free built-in service or a scoring service you hold a key for. Fraud systems judge a visitor by that address, so the score shows how a site is likely to see you. Each service draws its own lines, and our guide to a good IP fraud score explains them.

How do you read the results?

Three readings decide the test.

Is each exit on a home internet provider?

Read the Network column first. For a residential plan, every exit should sit on a home or mobile internet provider. An exit on a network flagged as hosting is a datacenter exit, whatever the plan is called. Websites read the same network record when they decide whether a visitor is a proxy.

To check a network name you do not know, ask the registry. IANA hands out address blocks to five regional registries, and they publish who holds each block. They answer over RDAP, a protocol that returns the record as JSON (RFC 9082). ARIN's server forwards each query to the right registry, so one command works for any address:

# Ask the registry who holds an IP (any region: ARIN forwards the query)
curl -s https://rdap.arin.net/registry/ip/75.75.75.75
# In the JSON, read "name" (the network) and the entity whose role is "registrant"

We ran the same lookup as a short script on two known addresses, one from AWS's own list and one of Comcast's:

Run on our own machine on October 23, 2026, against ARIN's RDAP server and AWS's ip-ranges.json and geo-ip-feed.csv of the same day. Both are public lookups of two known addresses, and no proxy is involved.

The first address belongs to Amazon's cloud, in the list AWS publishes so that "you can identify traffic from AWS". The second belongs to Comcast, a home internet provider, and that is the shape a residential exit should have. Our IP lookup shows the network and the operator of any address in a browser.

AWS is one of several clouds that publish such a list. We downloaded the lists of six clouds on October 23, 2026, and counted the unique IPv4 addresses in each.

IPv4 addresses that six clouds publish as their own
AWSip-ranges.json
102M
Microsoft Azurethe AzureCloud service tag
54.4M
Google Cloudcloud.json
19.4M
Oracle Cloudpublic_​ip_​ranges.json
4.3M
DigitalOceanits geofeed
3.1M
Akamai (Linode)its geofeed
1.4M
Together the six lists hold 184.6 million addresses, 4.30 percent of all IPv4 addresses.
Source: Each cloud's own published list, downloaded October 23, 2026; unique IPv4 addresses after merging overlapping prefixes, our count · hproxy.comProxy.

An exit inside any of these lists is a datacenter exit, whatever the plan calls it. A missing match proves less. AWS writes that it does not publish the ranges of all its services, and most hosting companies publish no list at all, so the Network column and the registry stay the main check.

Free proxy lists show the datacenter pattern at its extreme. In our free proxy data study, the largest network among live free proxies by far was Amazon's cloud, and are free proxies safe covers what else comes with them. One product passes the network check by design. An ISP proxy uses an address registered to an internet provider, served from a machine in a data center, so if ISP proxies are what you buy, that is the product. Our guide to ISP and residential proxies explains the split.

Do the exits land where you asked?

Count how many exits landed in the country or city you asked for. On a plan that sells that targeting, they all should.

Take the place from the checker's Location column, and leave the registry's address aside. The registry records where the holder is registered, which can sit far from where an address is used. Our capture shows the gap. The registry places the AWS block with an Amazon company in Herndon, Virginia, while AWS's own location file puts the same block in Houston, Texas. Networks publish such files as geofeeds (RFC 8805), a format that, in its own words, "only allows specifying coarse-level location".

Does it work on your own target?

This reading decides the purchase. A proxy can pass every column and still be refused by the site you care about. Send 20 to 50 real requests through the proxies to the page you plan to use, the way your tool will send them, and count how many return the page you wanted.

Note the speed on your target as well. The checker's Speed column measures a round trip through the proxy to our checker, and your job lives on the round trip to your target. Count the different networks too: twenty exits from one network make a small pool for your purpose, whatever the homepage says.

Which red flags should end the test?

  • Exits on hosting networks under a residential label.
  • Exits outside the country or city you paid to target.
  • A transparent grade, which means the proxy forwards your real IP.
  • Green checks, and failures on your own target, again and again.
  • Terms with no refund for unused traffic, or pages that disagree about refunds.
  • Traffic that lapses before you can finish testing it.
  • A price you can only learn on a sales call.

What goes on your checklist?

Five checks cover the terms and the test. Each note says what a pass looks like.

Five checks before you buy more than a test
  • Did you read the refund rule first?a window for unused traffic, in writing
  • Will the traffic last?long enough to finish the job, or no expiry at all
  • Is each exit on a home provider?the Network column, then the registry
  • Do the exits land where you asked?the Location column against your targeting
  • Does your own target answer?real requests, counted, at a speed your job can use
Source: HProxy, from our checkers' own report fields, the registry, and our records of sellers' terms of September and October 2026 · hproxy.comProxy.

Where are sellers compared side by side?

This guide tests one seller at a time. To choose between sellers, start with our pick of the best residential proxies, the price study of what 10 GB costs at 57 sellers, and the cheapest residential proxies with their conditions. What proxies actually cost is the benchmark for any price you are quoted.

Can you vet HProxy the same way?

Yes. Top up a small amount on our residential proxies page and run the same test. Residential costs $0.50/GB for one GB, falling to $0.44/GB at 2,000 GB+. Our terms describe "the 14-day money-back guarantee on unused balance", and the refund policy sets out the rest. They also say that purchased traffic "has no scheduled expiry date", and they list the uses we forbid. Our compliance page explains when we may ask for an ID. Other answers are on our FAQ.

Frequently asked questions

How do I test a proxy provider before I buy more?

Read its refund and expiry terms, then top up the smallest amount it sells. Paste 10 to 20 of its proxies into an online proxy checker, or check them from your own connection with a checker app. Read the network, location and anonymity of each exit, then send real requests through them to the site you plan to use.

How can I tell if residential proxies are actually datacenter?

Read the network of each exit. A proxy checker names the internet provider behind the address and flags networks that belong to hosting companies. A registry lookup over RDAP confirms it: a home internet provider such as Comcast holds a residential exit, and a cloud or hosting company holds a datacenter one.

What should I read before I top up?

Four terms: the refund rule, how long traffic lasts, when the seller asks for an ID, and which uses it forbids. In our records of 39 sellers' refund terms from September and October 2026, 18 allowed no refund for unused traffic, so read the rule before you pay.

Can I check proxies that only work from my own IP?

Yes, with a checker that runs on your own machine. An online checker connects from its own servers, so a proxy locked to your IP fails there. Our free proxy checker app runs the checks from your connection.

What does a good result look like?

Every exit works, sits on a home or mobile internet provider, lands in the country you asked for and grades elite or anonymous. Your own target decides the rest: real requests to the site you plan to use should succeed at a rate and a speed your job can live with.

Are cheap proxy providers always worse?

Price alone tells you little in either direction. Run the same test on a cheap seller and an expensive one: who holds the exits, where they land, how they do on your target, and what the terms say. The seller that passes at the volume you need is the better buy.

Get proxies that are alive right now

Our free proxy list re-checks every exit every few minutes across 100+ countries, with a live last-checked time, so you copy IPs that worked moments ago, not a stale text dump. When the location has to survive a real check, the paid network holds up.

129M+ proxy checks run · 100+ countries · HTTP / HTTPS / SOCKS · re-checked every few minutes · no signup

HProxy.

Honest guides and comparisons on proxies, scraping and staying unblocked, from the team that runs the network.

RSS feed