Per GB, Per IP or Unlimited? Proxy Pricing Models Explained

Why pricing models matter more than prices

Two providers can both look "cheap" and differ by 10x for your workload, simply because they bill in different units. Before comparing headline prices, understand what you're actually paying for.

The five common pricing models

1. Per gigabyte (bandwidth-based)

You pay for the data transferred through the proxy, in both directions. This is standard for rotating residential proxies and common for mobile proxies. You get access to the whole pool, and the number of IPs you use is effectively unlimited.

Best for: high-diversity scraping of lightweight pages and APIs. Watch out for: heavy pages, browser automation, video, and traffic that expires at the end of the month.

2. Per IP (per proxy, per month)

You pay a fixed price for each IP, usually with unlimited or generous bandwidth. Standard for datacenter proxies and ISP proxies.

Best for: high-bandwidth tasks on a predictable number of IPs, account management, sneakers. Watch out for: paying for idle IPs, and IPs that get banned on your target.

3. Per port or per concurrent connection

Some providers sell access by the number of simultaneous connections ("threads" or "ports") to a rotating pool, with unlimited bandwidth. Common with some rotating datacenter and backconnect services, and with dedicated mobile proxy ports.

Best for: heavy, steady workloads where you can keep every port busy. Watch out for: concurrency caps that throttle bursts.

4. Per request or per successful result

Scraping APIs and web unblockers often bill per request, sometimes only for successful ones, with higher prices for JavaScript rendering or protected targets. See scraping API vs proxies and web unblocker proxies.

Best for: hard targets where you'd otherwise burn lots of bandwidth on failures. Watch out for: credit multipliers for "premium" domains.

5. Unlimited bandwidth plans

Some residential and datacenter plans offer unlimited data for a time period, priced by speed, concurrency or server. See unlimited bandwidth proxies.

Estimating your bandwidth

For per-GB plans, estimate before you buy:

1. Open the target page in Chrome DevTools → Network, with cache disabled.
2. Note the "transferred" total at the bottom.
3. Multiply by pages per month.
4. Add 20% for retries, redirects and TLS overhead.

Typical transfer sizes (very rough):

Example: 200,000 product pages per month as raw HTML at 150 KB each is about 30 GB. The same pages in a headless browser at 3 MB each is about 600 GB. Same data, 20x the cost on a per-GB plan. That's why Puppeteer proxies setups should block images and fonts.

Which model is cheapest?

Always compare cost per successful request, not cost per unit. Our proxy pricing guide has a full worked example, and cheap proxies lists the biggest savings.

Hidden costs to check

• Traffic expiry. Many monthly GB plans expire unused traffic. Non-expiring and pay-as-you-go plans avoid this.
• Overage rates. What happens when you exceed your plan?
• Targeting surcharges. City, ASN or premium pools.
• Minimum commitments. Enterprise plans often require monthly minimums.
• Charging for failed requests. Per-GB plans charge for block pages too.

Where the bytes go

Per-GB billing typically counts traffic in both directions through the proxy, including overhead. For a single HTTPS page fetch, that includes:

• The CONNECT request and proxy response.
• The TLS handshake (certificates can be several KB).
• Request headers (cookies can be large).
• The response headers and body (compressed if your client allows it).
• Any redirects, which repeat much of the above.

For tiny API calls, overhead can be a large share of the total. Reusing connections within a sticky session reduces handshake overhead.

Measuring your real usage

Before buying a big plan, measure:

import requests
proxy = "http://USER:PASS@gate.example.com:7777"
r = requests.get("https://www.example.com/product/123",
                 proxies={"http": proxy, "https": proxy},
                 headers={"Accept-Encoding": "gzip, br"}, timeout=30)
print("decompressed bytes:", len(r.content))
print("compressed bytes (approx):", r.headers.get("Content-Length"))

Your provider's dashboard is the source of truth: run 100 requests on a test sub-user and read the usage it reports, then divide.

Pricing model by proxy type at a glance

A worked comparison

You run 30 browser sessions all day on a protected site; each session transfers about 200 MB per day with assets blocked.

• Per-GB residential: 30 × 0.2 GB × 30 days ≈ 180 GB/month. At $3/GB, about $540.
• ISP per IP: 30 IPs × about $2.50 ≈ $75/month with unlimited bandwidth.

Same workload, roughly seven times cheaper on per-IP pricing, provided ISP proxies perform on the target. That's why matching pricing model to traffic shape matters. See ISP proxies and proxy pricing guide.

How providers count a gigabyte

Two plans at the same price per GB can bill quite differently, because providers measure traffic in different ways. Before comparing, find out:

• Which directions count. Many providers bill upload and download together. For scrapers that's mostly download, but form-heavy automation and file uploads add up.
• Whether overhead counts. TLS handshakes, HTTP headers and retries all move bytes through the gateway. With per-request rotation, each new connection repeats the TLS handshake, which adds a few kilobytes every time.
• Whether failed requests count. A blocked response or a CAPTCHA page still transfers data. Some providers bill it like any other traffic; a few exclude certain errors. Ask.
• What a gigabyte is. Some dashboards use 1,000 MB, some 1,024 MiB. The difference is only about 2.4%, but it matters when you're comparing close offers.
• How usage is rounded and when it expires. Monthly traffic that resets unused, or rounding at the account level, changes the effective rate.

Measure your own mix rather than trusting averages. Compare the provider dashboard against your client's byte counts for a test day, and investigate any gap bigger than a few percent.

The fastest savings usually come from transferring less, not from a cheaper rate: block images and fonts in browsers, request compressed responses and avoid fetching pages you don't need. See cheap proxies for more ways to cut bandwidth, pay-as-you-go proxies for traffic that doesn't expire, and unlimited bandwidth proxies for when metering stops making sense.