Pay per GB proxies
How gigabytes are counted when you pay per GB, the 低价线路 and 优质线路 prices, how to estimate your volume and when leasing an IP costs less.
Estimate monthly GB from requests, size per request and retries, then check the number with a small paid test.
Per-GB billing counts the bytes that pass through the proxy, in both directions: the page, its scripts, its images, and every request you repeat after a failure. Only requests you actually send are counted, so the same URL fetched with plain HTTP costs far less than the same URL opened in a full browser.
For scale, the HTTP Archive Web Almanac measured a median page weight of 2,652 KB on desktop and 2,311 KB on mobile in October 2024, spread over a median of 71 requests on desktop. Images (1,054 KB) and JavaScript (620 KB) are the biggest parts. A browser that loads all of that spends the bytes; an HTTP client that fetches only the HTML or JSON does not.
| Item | Requests a month | Size of one request | GB a month, with 10% retries | Cost at US$1.50/GB (Budget line) |
|---|---|---|---|---|
| A. Price checks, plain HTTP | 500 products × 4 checks a day × 30 days = 60,000 | 150 KB | 60,000 × 150 KB = 9.0 GB; × 1.10 = 9.9 GB | 9.9 × 1.50 = US$14.85 |
| B. Browsing profiles in a browser | 20 accounts × 3 sessions a day × 30 days = 1,800 sessions | 3 pages × 2.65 MB ≈ 8 MB per session | 1,800 × 8 MB = 14.4 GB; × 1.10 = 15.84 GB | 15.84 × 1.50 = US$23.76 |
| C. API-style requests | 200,000 | 20 KB | 200,000 × 20 KB = 4.0 GB; × 1.10 = 4.4 GB | 4.4 × 1.50 = US$6.60 |
Sizes in A and C are assumptions to be replaced by your own measurement. In B, 2.65 MB is the HTTP Archive median desktop page (2,652 KB) used as a starting point. 1 GB = 1,000 MB here. The 10% retry margin is an assumption. Price: Proxy302 Budget line per-GB price, checked 10 November 2026.
Example A is a scraper that reads product pages with an HTTP client and never loads images or scripts. Example B opens each profile in a real browser, so most bytes are images and scripts, and the monthly total is larger than A even though the number of visits is much smaller. Example C sends small JSON requests, so a large number of requests still costs only a few GB.
The lesson is that request count alone tells you little: size per request decides the bill. If B ran on the Premium line at US$5 per GB, 15.84 GB would cost 15.84 × 5 = US$79.20, so the line you choose matters as much as the volume.
import requests
from urllib.parse import quote
user = quote("USERNAME", safe="")
password = quote("PASSWORD", safe="")
proxy = f"http://{user}:{password}@proxy.proxy302.com:2222"
proxies = {"http": proxy, "https": proxy}
urls = ["https://example.com/"] * 20 # replace with 20 real URLs of your job
total = 0
for url in urls:
r = requests.get(url, proxies=proxies, timeout=(5, 20))
total += len(r.content)
average_kb = total / len(urls) / 1000
print(f"average body size: {average_kb:.1f} KB")
requests_per_month = 60000 # your own number
gb = requests_per_month * average_kb / 1_000_000
print(f"about {gb:.1f} GB a month before retries")Example only. Replace USERNAME and PASSWORD with the values from My assets, and the URLs with your own. len(r.content) counts the response body after decompression, so it understates what the proxy carried: treat the result as a lower bound.
Pick 20 real URLs from your job and run the script above, or your own tool. Write down the average size and how many requests you will send a month.
Multiply to get GB, then add a margin for retries and blocked responses. 10% is a starting guess; use your own failure rate once you have one.
Buy a small amount of traffic, run a day of the real job, and compare the traffic used with your estimate. Scale the monthly number from that day, not from the guess.
Rotating residential traffic starts at 1 GB, so a test costs about US$1.50 on the Budget line or US$5 on the Premium line. Purchased traffic has no expiry, which means a test purchase is not wasted if your real number turns out larger: what you do not use stays available. There is no free trial.
Concurrency does not change how many GB you need, only how fast you use them. How per-GB billing works in general is covered in the pay-per-GB guide.
Multiply monthly requests by the average size of one response, then add a margin for retries. A plain HTTP request is often tens to hundreds of KB; a full page in a browser is often a few MB.
Usually yes, because it downloads images, scripts and styles that an HTTP client skips. The HTTP Archive median desktop page was 2,652 KB in October 2024.
The bytes that pass through the proxy count, including repeated requests after a failure, so include a retry margin in the estimate.
1 GB of rotating residential traffic. There is no free trial, and purchased traffic does not expire.
Product observation · Confirmed on:
Product facts file of the site, checked 10 November 2026 against the console, the backend and the price table: rotating residential Budget and Premium per-GB prices and packages (1 PTC = US$1), no expiry of purchased traffic, protocols and ports, no free trial.
Third-party source · Confirmed on:
Read on 12 November 2026: median page weight measured in October 2024 is 2,652 KB on desktop and 2,311 KB on mobile; the median page makes 71 requests on desktop and 66 on mobile; the median desktop page using images requests 1,054 KB and 620 KB of JavaScript.
Medians over the whole web, not over any one site; a given site can be far lighter or heavier.
Create an account, top up once, then pick rotating residential, static ISP or datacenter IPs and have one running in your browser within minutes.