Per IP vs Per GB Proxy Pricing: Choosing a Billing Model
By the ProxyV4 Engineering Team · Updated:
Per-IP proxy pricing charges a fixed amount for each IP address per month, while per-GB pricing charges for the traffic you send and receive through a provider's network, usually drawing IPs from a shared pool. Per-IP plans tend to cost less when each IP carries a lot of traffic or must stay the same for weeks; per-GB plans tend to cost less when you need many different IPs for light, occasional requests. The way to decide is to estimate your monthly traffic, then compare it with the number of fixed IPs the job really needs.
Per IP vs per GB proxy pricing at a glance
The two models bill for different things, so they suit different jobs. The table compares them as they are commonly sold; individual providers vary, so read the plan terms.
| Question | Per-IP pricing | Per-GB pricing |
|---|---|---|
| What you pay for | Each IP address, per month or term | Traffic, billed per GB (how upload and download are counted varies by provider) |
| Which IPs you use | Fixed addresses assigned to you | Usually a shared pool, rotating per request or sticky for a while |
| Bandwidth | Often unmetered, but some per-IP plans cap it | The thing being metered |
| Bill predictability | The bandwidth part of the bill is fixed | Rises and falls with traffic, retries and page weight |
| Typical products | Datacenter and static ISP proxies | Rotating residential and mobile pools; some static ISP products |
| Best fit | Heavy or steady traffic, long sessions, fixed IPs | Many IPs or locations, light or bursty traffic |
What each model actually bills for
Neither model is cheaper in general. Each one prices a different scarce resource, and your job decides which resource you use more of.
Per-IP: you rent addresses
Under per-IP pricing, you pay for a set number of addresses for a term. If the plan is unmetered, the same price applies whether an IP moves a few megabytes or a lot of data that month, so the cost grows with the number of IPs you need, not with page sizes. The trade-off is that you pay for each IP whether you use it or not, and you have to manage them: which worker uses which IP and how fast each one sends requests.
Not every per-IP plan is unmetered. Some providers cap bandwidth per IP or apply a fair-use limit, so check before you assume traffic is free.
Per-GB: you rent traffic through a pool
Under per-GB pricing, you pay for data volume and usually get access to a large pool of IPs across many locations. That is valuable when a job needs breadth: checking search results from many cities, or sending a few requests each through many different addresses. You pay nothing extra for the size of the pool, and an idle month costs little.
The cost follows everything that crosses the meter: request and response headers, page bodies, images and scripts a browser loads, redirects, error pages and retries. Bills move with page weight and with how well your code avoids wasted requests.
Estimate your traffic before you compare prices
Price lists are only comparable once you know roughly how many GB a month your job moves. Measure real requests rather than guessing from page sizes in a browser, because compression and headers change the numbers.
Measure the bytes per request
The script below fetches a sample of URLs through a proxy and counts the bytes that actually cross the connection: request headers and body, response headers, and the response body as received, still compressed. It reads the body with r.raw.read(decode_content=False), because r.content would give the decompressed size and overstate metered traffic for gzip responses.
import requests
PROXY = "http://USER:[email protected]:8080"
proxies = {"http": PROXY, "https": PROXY}
# Replace with a sample of the URLs your job really fetches
URLS = [
"https://example.com/",
]
def header_bytes(headers):
# Each header line is "Name: value\r\n": name + value + 4 bytes
return sum(len(k) + len(v) + 4 for k, v in headers.items())
def transfer_bytes(url):
with requests.get(url, proxies=proxies, timeout=15, stream=True,
headers={"Accept-Encoding": "gzip, deflate"}) as r:
body = r.raw.read(decode_content=False) # body as received, still compressed
sent = header_bytes(r.request.headers) + len(r.request.body or b"")
received = header_bytes(r.headers) + len(body)
return sent + received
sizes = [transfer_bytes(u) for u in URLS]
avg_kb = sum(sizes) / len(sizes) / 1000
print(f"Average per request: {avg_kb:.1f} KB (TLS and TCP overhead not included)")The result excludes TLS handshakes, TCP/IP packet overhead and the proxy's CONNECT exchange, which providers may or may not count. It also covers only what your script fetches; a headless browser that loads images, fonts and scripts transfers far more per page.
Turn it into GB per month
Multiply requests by average size: pages per month x average KB per page / 1,000,000 = GB per month (using decimal units, which most price lists use). Then add a margin for retries, redirects and pages heavier than your sample.
With that figure you can price the per-GB option directly. For the per-IP option, the question is different: how many fixed IPs does the job need to stay within each target site's rate limits and keep sessions on stable addresses? Price that number of IPs, then compare the two totals.
Costs that do not show on the price list
A few factors decide the real bill under either model:
- Retries and failures. On metered plans, failed attempts and error pages are traffic too. Fixing a retry loop can cut a per-GB bill; on an unmetered per-IP plan the waste costs time, not money. Either way, when a site returns HTTP 429, honor
Retry-Afterand slow down. - Compression. Sending
Accept-Encoding: gzip(most libraries do by default) can shrink text responses considerably, which matters a great deal on a meter. - Browsers versus HTTP clients. Rendering pages in a headless browser pulls in every asset; blocking images or fonts you do not need saves metered traffic.
- Session needs. If a login or checkout flow must stay on one IP, pool-based plans need sticky-session settings that can expire mid-task, while a fixed IP simply stays put.
- Management work. Per-IP plans leave allocation and pacing to you. Per-GB pools handle IP selection for you, at the price of less control over which IP you get.
Which jobs are cheaper under each model
Per-IP pricing usually wins for steady, traffic-heavy work on a manageable number of IPs: downloading catalogs or documents, scheduled scraping of the same sites every day, running logged-in sessions for separate business accounts, and anything a partner must allow-list by IP. If you are close to the line, the deciding factor is often predictability: a fixed bandwidth cost is easier to budget than a meter.
Per-GB pricing usually wins for broad, light work: small requests spread across many locations, short projects, or jobs that need a different IP for each request. If your traffic estimate is small and your IP count would be large, a pool is the better buy.
Many teams use both, a handful of fixed IPs for sessions and heavy transfers and a metered pool for wide sampling.
Where ProxyV4 pricing fits
ProxyV4 sells per-IP only. Private datacenter IPv4 is available in the United States, Vietnam, Japan, South Korea, Singapore, the United Kingdom and Hong Kong; static ISP proxies are fixed IPs registered to consumer internet providers in the United States and Vietnam. Traffic on both is never metered or billed, each IP has one tenant for the term, and renewing before expiry keeps the same address.
Prices are per IP per month in USD, for 1, 3 or 12 months or a custom number of months, and the per-IP price falls as the order grows; see live prices. You can order a single IP, run the measurement above through it, and scale once the numbers work. Every paid proxy supports HTTP(S), and private IPv4 outside Vietnam and US static ISP proxies also support SOCKS5 with username and password login.
It is not the right fit if your job needs a large rotating pool, a new IP on every request or mobile carrier IPs: ProxyV4 has no per-GB plans, no rotating residential or mobile proxies, and rotating proxies are waitlist only. A proxy that fails for a technical reason, such as not connecting or showing the wrong location, is covered by the refund policy as a credit to your account balance.
Frequently asked questions
Are unmetered proxies really unlimited?
It depends on the provider, so read the terms for caps or fair-use limits. On ProxyV4 private IPv4 and static ISP proxies, traffic is never metered or billed.
Do per-GB plans count both upload and download?
Often they count both directions, but the rules vary by provider. Check how the plan measures traffic before using your estimate.
Is per-GB always cheaper for small scraping jobs?
Often, when the job is light and needs many IPs. If it needs a few fixed IPs for weeks, or pages are heavy, per-IP can cost less.
Do I keep the same IP when a per-IP plan renews?
With ProxyV4, yes: the IP stays the same while the proxy is active, and renewing before expiry keeps the same address.
What if a per-IP proxy does not work?
At ProxyV4, a proxy that fails for a technical reason, such as not connecting or showing the wrong location, is credited to your account balance under the refund policy.
Unmetered per-IP pricing for traffic-heavy jobs
ProxyV4 private IPv4 and static ISP proxies are priced per IP with traffic that is never metered. Order one IP, measure your real job through it, then scale.
Compare locations