IPv6 gives you addresses by the thousand at a very low unit price. It is the right tool when your targets support IPv6 and you need breadth — and the wrong one when they do not.
IPv4 addresses are scarce and individually priced; IPv6 addresses are abundant. A single /64 subnet contains more addresses than the entire IPv4 internet squared, so providers can hand you thousands cheaply.
The catch is reach. Many popular websites still do not serve over IPv6, so a request through an IPv6-only exit simply fails to reach them. Always check the target before buying — the IPv4 vs IPv6 guide shows how.
02
Where IPv6 proxies shine
IPv6 proxies are strongest on large platforms and CDNs that fully support IPv6 and rate-limit per address. Spreading requests across thousands of unique addresses keeps each one far below any per-IP threshold.
Because many networks apply rate limits to a whole /64 rather than one address, results vary by site. Start with a small pack, measure the success rate on your actual targets, then scale.
Use cases
Where it fits best
Wide request distribution
Keep per-IP request counts tiny by spreading traffic across thousands of addresses.
IPv6 connectivity testing
Verify that your own services, DNS and CDN behave correctly for IPv6 clients.
Public data collection on IPv6-ready sites
Fetch open data from platforms that serve IPv6 natively.
Setup
Set up IPv6 Proxy
01
Confirm the target supports IPv6
Check for AAAA records using the DNS lookup tool.
02
Order a small pack first
Start with 100 addresses to measure success rate on your real targets.
03
Configure HTTP/HTTPS access
Enter the supplied host, port and credentials into your client.
04
Scale up gradually
Move to larger packs once the success rate looks good.
An IPv6 address is 128 bits long. In a /64 the first 64 bits are the network prefix and the remaining 64 identify hosts, so a single /64 spans 2 to the 64th power addresses, about 18.4 quintillion. Your pack is a slice of that block, which is why large IPv6 packs can be priced so low: addresses are plentiful rather than scarce. In a typical setup the proxy software on the server picks one of your addresses as the source for each outgoing connection, so the destination sees an address from inside that /64 rather than the server’s own.
Destinations rarely treat every address as a separate visitor. Operators know that a household or office usually receives at least a whole /64, so abuse systems commonly count requests per prefix, often the /64 and sometimes a wider /56 or /48, instead of per single address. When that is the case, rotating through 15,000 addresses in one /64 looks to the target like one very busy client. Whether a given site keys on the full address or on the prefix is not something the proxy can change; you find out by measuring, as the sizing section below explains.
Prefix
Typical meaning
Size
/128
One single address
1 address
/64
The standard single network; the kind of block your pack is drawn from
2 to the 64th power addresses
/56
Often assigned to a residential or small-office customer
256 networks of /64 size
/48
Often assigned to an organization’s site
65,536 networks of /64 size
The number that matters is the prefix a target aggregates on, not the number of addresses you rent. Test that before you buy a large pack.
02
Is your target IPv6-ready? A pre-order test routine
An IPv6 proxy can only reach hosts that have an AAAA record and actually answer on IPv6. Start with the DNS lookup tool: query AAAA for the hostname, then repeat for every hostname the page pulls in — the main site, the API host, the asset or CDN host, and any login domain. A site can publish an AAAA record on its main hostname while a subdomain stays IPv4-only, and a single IPv4-only dependency is enough to break a scraper that needs that endpoint. Write down which hostnames pass; that list is your real compatibility matrix.
Do not treat a normal browser tab as proof. Browsers use Happy Eyeballs (RFC 8305): they try IPv6 and quietly fall back to IPv4 within a fraction of a second, so a page can load even when its IPv6 side is broken. A proxy with an IPv6-only exit has no such fallback, and the failure appears as a timeout or gateway error. Unless sales confirm otherwise, treat the exit as IPv6-only. Remember the two legs: your client reaches the proxy host over your own network, while the proxy reaches the target over IPv6; the host in your panel shows which family your side needs.
Run an AAAA lookup for the main hostname, the API host, the asset or CDN host and the login domain.
Open the page in a browser with the network panel visible and list every hostname it contacts.
Order the smallest pack, 100 IPs, and request one known IPv6-ready page through it.
Request one IPv4-only page as well, so you recognize the failure symptom later.
Record pass or fail per hostname before scaling the pack.
03
Sizing a pack from a rate-limit example
Sizing starts from the target’s limit, not from the pack catalog. Suppose, as an illustration, a site tolerates about 30 requests per minute from one address and you want 20 requests per second, which is 1,200 per minute. If the limit is truly per address, you need at least 40 addresses, and a 100-IP pack gives comfortable headroom at about 12 requests per minute each. If the limit is per /64 prefix, the same 1,200 requests per minute is 40 times the allowance and no pack size fixes that — you must slow down.
Measure the limit instead of guessing. Send the same modest burst, for example 60 requests in one minute, first from a single address and then spread over 20 addresses. If the second run draws far fewer 429 responses or challenge pages, the target counts per address and a larger pack helps. If both runs fail at the same point, it counts per prefix or per account, and the useful variable is your pacing. Record the result before choosing a size; packs are billed monthly only and the smallest costs 100 TRY (see the table), so this experiment is cheap.
Concurrency is a separate question from pack size. A 15,000-address pack does not mean 15,000 simultaneous connections; your client, its file-descriptor limit and the listed 100–500 Mbit port speed set the ceiling. As an illustration, 200 concurrent connections each moving 50 KB/s add up to 10 MB/s, roughly 80 Mbit/s, which already uses most of the lowest listed speed. More addresses thin out per-IP request rates; they do not raise throughput, so size the worker pool and the pack independently.
Pack
Monthly price (TRY, excl. taxes)
Per IP (TRY)
100 IPs
100
1.00
500 IPs
300
0.60
2,500 IPs
750
0.30
15,000 IPs
2,250
0.15
A bigger pack lowers the price per address but only helps if the target limits per address. Check that first.
04
Operating the pack: rotation, integration and responsible use
The pack is static, so rotation is a job for your client. The simplest scheme is random selection per new connection; the sturdier one is a per-target pool in which each address carries a cooldown. For example, after a 429 or a challenge page, retire that address for that target for 10 minutes (an illustrative figure) and pick another. Keep a counter per address and per target so you can see whether failures cluster on a few addresses, which points to reputation, or are spread evenly, which points to prefix-level limits or to your pacing.
Integration is ordinary HTTP proxying. Most clients accept a proxy URL of the form http://user:pass@host:port, curl takes it with the -x option, and Python requests takes it in its proxies setting; Node needs a proxy-agent package. Percent-encode special characters in the password, or the URL will parse wrongly. Because SOCKS5 is not confirmed for IPv6 plans, do not configure socks5:// URLs unless sales confirm it. For HTTPS targets the client sends a CONNECT request, so the proxy sees only the destination host and port, not the encrypted content.
Large address counts raise your responsibility rather than lower it. Spreading traffic thinly is a courtesy to the target, not a way to hide abuse, and it does not entitle you to ignore a site’s terms, robots.txt or published rate limits. Log the exit address, status code and timestamp of every failure, keep the log for your own audits, and stop or slow down when a target asks. Use the pack only for lawful work within the terms of the services you contact.
Before you order
Buyer’s checklist
0 / 6
Troubleshooting
When something does not work
The five symptoms we see most often, what usually causes them and the quickest fix.
Likely cause
The target has no AAAA record or does not answer on IPv6, and an IPv6-only exit has no IPv4 fallback.
Fix
Check AAAA with the DNS lookup tool and send that target through a datacenter or ISP proxy instead.
Likely cause
Some hostnames the page depends on, such as a CDN or API host, are IPv4-only.
Fix
List every hostname in the browser network panel, check AAAA for each, and fetch only IPv6-ready endpoints or split the job across proxy types.
Likely cause
The target counts requests per /64 prefix, per account or per fingerprint rather than per address.
Fix
Lower the aggregate request rate, add delays and jitter, and test whether one address and many addresses fail at the same point.
Likely cause
The listener speaks HTTP; SOCKS5 is not confirmed for IPv6 plans.
Fix
Switch the client to an http:// proxy URL, or ask sales whether SOCKS5 can be provided for your pack.
Likely cause
The echo service may not be reachable over IPv6, or your client is bypassing the proxy.
Fix
Use an echo endpoint that publishes AAAA records, confirm the client really uses the proxy, and compare with the proxy checker result.
Glossary
Terms you will meet
AAAA record
The DNS record type that maps a hostname to an IPv6 address, the counterpart of the A record for IPv4. No AAAA record usually means the host cannot be reached over IPv6.
/64 prefix
An IPv6 network whose first 64 bits are fixed, leaving 64 bits for hosts. It is the standard size of a single subnet and the kind of block your pack is drawn from.
Dual stack
A host or network that runs IPv4 and IPv6 side by side. A site can be dual stack on its main hostname while a subdomain remains IPv4-only.
Happy Eyeballs
The algorithm in RFC 8305 that lets a client race IPv6 and IPv4 connection attempts and use whichever succeeds first. It explains why a site can seem fine in a browser while an IPv6-only path fails.
Prefix-based rate limiting
Counting requests per IPv6 prefix, commonly a /64, instead of per single address. It limits the benefit of rotating inside one subnet.
Source address selection
The process by which a host or proxy chooses which of its local addresses to use for an outgoing connection. In an IPv6 proxy this is what makes each request appear to come from a different address in the pack.
FAQ
IPv6 Proxy — frequently asked questions
Will IPv6 proxies work on every website?
No. They only reach sites that are reachable over IPv6. Check the target’s AAAA record first, and keep an IPv4 proxy for the rest.
Are SOCKS5 proxies available for IPv6?
The provider lists HTTP and HTTPS for IPv6 plans. SOCKS5 is not confirmed, so we do not offer it as a feature.
What does “/64 subnet” mean?
It is the size of the address block your IPs come from. A /64 contains a vast number of addresses; your pack uses a slice of it.
Where are the IPs located?
The order page does not show a location selector. If location matters, ask sales before ordering.
Should I buy IPv6 or datacenter IPv4?
IPv6 when your targets support it and you need many addresses cheaply; datacenter IPv4 when you need universal reach.
Does each IPv6 address come with its own port?
The order page does not describe how the delivered pack is laid out, so we cannot say here whether it is a single endpoint or one entry per address. Check the delivery in your customer panel after ordering, and ask sales on WhatsApp first if your tool needs a specific layout.
Do I need IPv6 on my own network to use these proxies?
Not necessarily. The connection from your machine to the proxy and the connection from the proxy to the target are separate legs, and only the second is IPv6. If the host you are given is reachable over IPv4, your own network does not need IPv6; if it is an IPv6 address, it does. Test from your own machine before building on it.
Are IPv6 addresses trusted more or less than IPv4 addresses?
There is no universal answer. The provider lists the addresses as previously unused, so they carry no history of their own, but every address in the block shares the surrounding /64, and some systems score the whole prefix as one client. The IP information tool shows RDAP registry data for an address, which helps you see the registered range around it. Measure on your own targets.
How do I move from the 100-IP pack to a larger one?
Packs are separate monthly plans, and no upgrade path is documented on this site. The straightforward route is to order the larger pack from its own order page; for anything else, such as changing size mid-period, ask sales. Refund and replacement terms are not published.
Can I test an IPv6 proxy for free before buying?
Not a pack you have not purchased. The free tools help on the target side: the DNS lookup tool shows whether a hostname has AAAA records, and the proxy checker can test your entries once you have them. The 100-IP pack at 100 TRY per month is the lowest-cost real test, and refund terms are not published, so ask sales first if that matters.