Skip to the answer
Proxy Costsobserved 2026-07-29

Proxy pricing terms

These are operational definitions used by this site and its machine exports. Unknown never inherits the most convenient meaning, and a product keeps the unit in which the provider sells it.

Terms with their own page

Each of these is explained on its own page with a real plan from our dataset where the term changes the price, named and sourced.

Minimum checkout cost
The smallest amount you can actually pay a provider to start using its proxies. It is not the advertised rate per GB multiplied by the traffic you want. Advertised rates usually belong to a volume tier, and you cannot buy a large tier's rate at a small tier's volume.
Effective cost per used GB
Checkout cost divided by the traffic you will actually use, rather than by the traffic the plan includes. If a plan sells 100 GB and you use 40, your effective cost is two and a half times the advertised rate.
Bandwidth expiry
How long unused traffic stays usable. Traffic that expires in 30 days is a cheaper product than traffic that never expires, and providers price it that way. Where the same plan is sold at several expiry windows, each one has its own price.
Rollover
Whether unused traffic carries into the next billing period. Rollover is the difference between paying for what you use and paying for what you forecast.
Bandwidth pooling
Whether one balance of traffic can be shared across sub-users, proxies or plans, or whether each is metered separately. Pooling decides whether you buy one large allowance or several small ones, and several small ones almost always cost more per GB.
Overage
What you pay per GB after the included traffic runs out. An overage rate well above the plan rate turns an under-sized plan into an expensive one.
Concurrency limit
How many requests or sessions a plan lets you run at once. A plan you cannot run your thread count on is not cheap for your job, it is unusable for it. This is where inexpensive plans are usually made inexpensive.
Thread limit
A per-plan cap on parallel connections, stated separately from concurrency by some providers. Worth checking against the scraper you actually run before comparing prices at all.
Rotating proxy
An address that changes on every request or at a fixed interval. Suited to work where each request should look unrelated to the last.
Sticky session
An address held for a set period, from a minute to several hours depending on the provider. Necessary for anything that carries a login or a cart across requests, and the maximum duration varies enough between providers to decide the choice.
ISP or static residential proxy
A residentially registered address hosted in a datacenter. It combines datacenter speed with residential registration, and it is usually sold per IP per month rather than per GB, which makes it hard to compare against per-GB plans without doing the arithmetic properly.
Geo premium
An extra charge for targeting a specific country, state, city or network. Some providers include country targeting and charge for city or ASN targeting, so the advertised rate is only the rate for untargeted traffic.
Unlimited bandwidth plan
A plan sold without a traffic meter. The cost is moved somewhere else, usually into a concurrency cap, a speed cap, or a per-port fee. Unlimited describes the meter, never the capacity.

More terms with their own page

These have a page but no entry in the list above, so they are named here rather than left reachable only from the site index.

Defined here for now

These are accurate but do not yet have a page of their own. They are the queue for the next writing pass, and saying so is more useful than a stub that pretends otherwise.

Success rate
The share of requests that return a usable response. It is measured by whoever ran the test, on their targets, at their moment, so it belongs next to the tester's name and date rather than inside a price. We run no tests of our own.
Cost per successful GB
Effective cost per used GB divided by the success rate. It is the only figure that lets a cheap, unreliable plan be compared honestly against an expensive, reliable one, and it is only as good as the success rate it is given.
Not published
The provider does not list a public price for this plan. It is a fact about the provider, not a gap in the data, so the row stays visible with a link to the provider's own page instead of an estimate.

Last materially reviewed: