Loading page content.
A routed IPv6 block of your own, at a fraction of IPv4 pricing. Rotation becomes a socket option instead of a purchase.
Starting price
$0.30/IP/wk
per IP, week, from 10 IPs. Falls to $0.15/IP/wk at 10,000 IPs.
Every account starts with 50 MB of free residential bandwidth. No card is required to test the network.
How it works
IPv6 changes the economics of address diversity. Where IPv4 addresses are scarce and priced accordingly, a single delegation hands you tens of thousands of addresses — so the question stops being how many you can afford and becomes which one this request should use.
Where a target answers over IPv6, a /48 hands you 65,536 addresses to spread requests across for a fraction of what the same footprint costs in IPv4. Blocks are routed to one tenant, so you rotate through them client-side at whatever rate you like without asking us. Sold in steps of ten addresses.
You buy addresses in tens from our own /32 allocations. Above 4,096 addresses the block is delegated as a routed /48, which you may subnet however you like.
Because the whole block routes to you, selecting a source address is a local operation — a bind call, a curl --interface flag, an httpx transport option. It costs nothing, needs no gateway round trip, and has no rotation quota attached.
The block is routed to one tenant. No other customer draws from your addresses, so the reputation of every address in it is a product of your own traffic.
Blocks sit on 10 Gbps ports with no bandwidth counter. The weekly rate per address is the whole cost of the product.
The honest limitation, and it is a real one: the target has to answer over IPv6. Much of the commercial web still publishes no AAAA record, and some sites that do treat IPv6 traffic more suspiciously than IPv4 because so little legitimate consumer traffic arrives that way. Check your targets before you buy a block.
Volume ladder
Tiers are a step function, not interpolation: the greatest break at or below your quantity wins and every unit bills at that rate. They apply automatically at checkout — there is no code to enter and nothing to negotiate.
| Quantity | Unit price | Saving | Order total | Order at this tier |
|---|---|---|---|---|
| 10 IPs and above | $0.30/IP/wk | Entry rate | $3.00 | Order |
| 100 IPs and above | $0.27/IP/wk | Save 10% | $27.00 | Order |
| 500 IPs and aboveMost popular | $0.24/IP/wk | Save 20% | $120.00 | Order |
| 1,000 IPs and above | $0.21/IP/wk | Save 30% | $210.00 | Order |
| 5,000 IPs and above | $0.18/IP/wk | Save 40% | $900.00 | Order |
| 10,000 IPs and aboveBest value | $0.15/IP/wk | Save 50% | $1,500.00 | Order |
Totals are what checkout charges for exactly that quantity. Buying past a break lowers the rate on every unit, not only the ones above it. Price a custom quantity in the calculator.
Specification
Everything a reader needs to decide whether this product fits their stack, on one screen and in one table.
| Property | Value |
|---|---|
| Rotation | Client-side, unlimited, across every address in the block, at no cost per change |
| Sticky duration | Entirely your choice — you own source-address selection, so an address is held for exactly as long as you keep using it |
| Targeting | Region: United States, Germany, Netherlands, Singapore and Japan |
| Protocols | HTTP and HTTPS via CONNECT on 8080 and 8443, SOCKS5 on 1080. IPv6 destinations natively, dual-stack on request |
| Concurrency | Unlimited across the block on a 10 Gbps port; in practice one worker per address is the natural shape |
| Bandwidth policy | Unmetered transfer for the life of the block |
| Authentication | Username and password, or an IP whitelist on the block's endpoint |
| Order size | 10 IPs to 100,000 IPs per order, in steps of 10 |
Integration
There are two ways to use a block, and they suit different clients. Bind a source address locally when your client supports it; go through the block's proxy endpoint when it does not.
fp_a41c9d2001:db8:4a2::/64
The block routed to your account. Every address inside it is yours to source from.
fp_a41c9d
The credential id, used when you connect through the block's proxy endpoint rather than binding locally.
Local bind
curl --interface, httpx local_address, or a bind() call — the cheapest rotation available anywhere in this catalogue.
Delegation
Above 4,096 addresses the block arrives as a routed /48, which you may subnet across hosts yourself.
# The whole block routes to your host, so rotation is a source-address
# choice your client makes. Two requests, two addresses, no API call.
curl --interface 2001:db8:4a2::1f -s https://api.fleetproxy.com/v1/ip
curl --interface 2001:db8:4a2::20 -s https://api.fleetproxy.com/v1/ipTargeting
Region is chosen when the block is allocated. Everything below that — which address, how often it changes, how they map to workers — is yours to decide in code.
| Control | How it is set | Notes |
|---|---|---|
| Region | Checkout selection | United States, Germany, Netherlands, Singapore or Japan. |
| Address | Local bind | Any address in the block, chosen per request or per worker. |
| Block size | Sold in tens | From 10 addresses upward; a routed /48 is delegated above 4,096. |
| Dual stack | On request | For targets that publish no AAAA record, pair the block with an IPv4 lease. |
IPv6 blocks are announced from the regions below. The counts are the wider datacenter inventory in each market.
See every country and city with live inventory — counts are the addresses published for that market, not a global total repeated on every page.
Use cases
Four workloads where this product is the right answer rather than a workable one.
Large properties with full AAAA coverage — search infrastructure, major CDNs, modern APIs — can be crawled at a per-address cost that makes IPv4 look extravagant. One address per worker, thousands of workers, one weekly line item.
Checking how a receiving system treats mail and API traffic from a clean, dedicated block, and how quickly reputation accrues on addresses nobody else has used.
Per-IP rate limits are the usual constraint on polling a public API at volume. A block of addresses converts that limit into a scheduling problem rather than a spending one.
Measuring how a target's defences respond to address spread — how many requests per address, over what window, before behaviour changes. Free rotation is what makes the experiment affordable.
Browse every use case for the full breakdown, including the products that suit each one.
Compare
The two nearest products in the catalogue, and the conditions under which each of them is the better purchase. Buying the wrong class costs more than buying the wrong volume.
$2.00/IP/wk
Choose datacenter IPv4 instead if your targets publish no AAAA record, which is still most of the commercial web. Test with a single curl against each target before you size a block: an IPv6-only plan against an IPv4-only site simply does not connect.
$1.20/GB
Choose rotating datacenter instead if you want IPv4 breadth without owning the address-selection logic. The pool does the rotation for you and bills per gigabyte, which suits bursty work better than a fixed weekly block.
Compare all eight proxy types on identity, rotation, targeting and billing, or price them side by side.
Questions
Pulled from the help center and filtered to the questions this product actually raises.
Residential IPs are addresses assigned by consumer ISPs to real subscriber connections, so targets see an ordinary visitor; they cost more and add 200–900ms of latency. Datacenter IPs come from hosting providers, are roughly twenty times cheaper per gigabyte and much faster, but are trivially identifiable and filtered by any site that cares. Mobile IPs are carrier-assigned 4G/5G addresses shared by thousands of subscribers, which makes them the hardest to block and the most expensive. Start with datacenter on undefended targets and residential everywhere else.
Roughly 30 million residential addresses across 195 countries, plus dedicated static residential, mobile and datacenter inventory. Coverage is not uniform: the locations pages show live IP counts per country and city so you can check a specific market before ordering rather than after. Anything under 500 IPs in a pool is flagged, because a pool that small will repeat addresses inside a large crawl.
No. Create an account and claim the 50 MB free trial of rotating residential bandwidth — no payment method, no call. That is enough to run a few hundred requests against your real target and measure the success rate before you spend anything.
No. We meter volume — bytes up, bytes down, request counts, success and failure counts, mean latency and exit country, aggregated per plan per hour. We do not record destination hostnames, URLs, headers or traffic content, and the usage tables have no column capable of holding them. Connection metadata such as your account's login IP is retained as described in the privacy policy.
No. Purchased bandwidth stays on the account for as long as the account is open and in good standing. Most of the industry expires unused traffic at the end of the billing period, which turns a bursty workload into 30–40% waste; we do not, and that is a policy commitment rather than a promotion.
Within 24 hours of purchase with under 100 MB consumed, we refund in full with no questions. Within 7 days with under 10% consumed, we refund the unused portion less irrecoverable processing fees. Beyond that the purchase is final, but the bandwidth never expires, so it is waiting whenever you come back. Use the free trial first — it exists to answer the compatibility question before money is involved.
Something not covered here? Read the full help center or ask the support team.
Start on 50 MB of free residential bandwidth, then buy from $0.30/IP/wk. Volume tiers apply automatically, and there is no contract to sign.
No card required for the trial. Cancel or downgrade at any time.