Bulk retrieval
High-volume reads against targets that do not score the range a request arrives from. The cheapest byte on the network.
The cheapest egress on the network, metered the same way as everything else. Per gigabyte, through one gateway, with no per-IP subscription and no list to maintain.
A datacenter proxy exits through an address registered to a hosting provider rather than to a consumer internet line. The address belongs to a data center, and anyone who looks it up can see that. There is no pretending otherwise, and no reason to.
That is a disadvantage on targets that score the range a request arrives from, and an advantage everywhere else. A datacenter address sits on a commercial link with commercial capacity behind it. No home router, no shared consumer uplink and no radio stands between your request and the response.
Buy datacenter when the target does not care where the request came from. Most targets do not. The ones that do are the reason residential exists, and the two are meant to be used together rather than argued about.
Datacenter traffic is $0.99 per gigabyte. Nothing is charged per address, per port, per seat or per month.
Datacenter proxies are usually sold as a list: a count of addresses at a monthly rate each. That model bills for capacity rather than use. An idle address costs exactly what a saturated one costs, so the pricing question becomes how many addresses to lease — a question you cannot answer before you have run the job.
Per-gigabyte billing removes the guess. You buy traffic, you spend traffic, and what you do not spend stays on the balance because credits carry over indefinitely. A week with no crawl costs nothing. A week that doubles costs twice as much and nothing more, with no plan to upgrade and no addresses to provision.
Two details worth knowing before you model it. The meter counts every byte on the wire — request headers, TLS negotiation, redirects and the responses you discarded, not just the payload you kept. And traffic routed through Target Filters, meaning state, city, ZIP or ASN selection, bills at twice the standard rate; country selection and ASN exclusion are included at the base rate. Both are laid out in what providers bill you for.
Every plan carries a 7-day money-back guarantee, and there is no minimum term.
Datacenter is the one tier issued randomized across the pool, with no subnet blocks. It is the row in the table most buyers skip, and it is the one that decides how long the addresses last.
Reputation on hosting ranges travels by neighborhood. A contiguous block leased to a single buyer behaves like a single identity: the moment one address in it is flagged, the ones on either side are already suspicious, because they demonstrably belong to the same customer. Anti-abuse systems widen from an address to its block precisely because blocks are how the addresses were sold.
Randomized allocation breaks that inference. The address serving your next request has no registration relationship to the address that served the last one, so a flag does not propagate to everything you are about to use. It is not invisibility. It is the difference between losing one exit and losing 256.
Rotating and sticky sessions are not part of the datacenter tier. There is no session window to hold and no sticky port to bind. If a workflow has to keep the same exit across a login, a cart or a multi-step form, this is the wrong pool and no configuration will change that.
That work belongs on residential at $1.99 per GB or mobile at $3.99 per GB, where ports 10000 to 20000 hold one address for a window between 1 and 120 minutes. The mechanics are in connecting to the gateway, and the choice between the two modes is covered in rotating vs sticky sessions.
Targeting is not a limitation here. Country selection and ASN exclusion are included at the base rate on this tier, and state, city, ZIP and ASN selection are available as Target Filters at twice the rate — the same controls the other two pools get. See targeting.
One table, three egress types, read line by line. Every value below is the same value the pricing table renders.
| Compare | DatacenterThis page$0.99/ GB | $1.99/ GB | $3.99/ GB |
|---|---|---|---|
| Address pool | Datacenter | 90M+ residential | 5G / 4G / 3G / LTE |
| Countries | 195 | 195 | 195 |
| Country targeting | |||
| State / city / ZIP / ASN targeting | |||
| Rotating & sticky sessions | |||
| Randomized, no subnet blocks | |||
| Pay per traffic, no expiry | |||
| Uptime | 99.9% | 99.9% | 99.9% |
| 24/7 support |
Read it as a routing decision, not a ranking. Datacenter is the cheapest byte and the only randomized pool. Residential adds ISP addresses and session control. Mobile adds carrier egress. All three rates sit on pricing.
High-volume reads against targets that do not score the range a request arrives from. The cheapest byte on the network.
Endpoints that publish deliberately. Nothing is gained by paying residential rates to read a documented feed.
Reach your own environments from outside your own network, without routing test traffic through the office range.
Give a pipeline a public egress that is not the shared address of whichever cloud region it happened to start in.
Large objects where the cost per gigabyte dominates and the target has no interest in who is fetching them.
Split by target, not by habit. Datacenter carries what it can and residential carries the rest.
Credentials, the first request and where usage is reported are all in getting started.
A readable slice of the datacenter footprint. The rest is counted, not listed.
Why one hostname replaces a spreadsheet of addresses, and what changes in your code.
Headers, TLS, redirects and failed responses all move bytes. Here is the full list.
What to strip, cache and skip before a request ever reaches the gateway.
Turn a page count into a gigabyte count before you buy the traffic.