The proxy industry has settled on three pricing models: per-GB (pay for bandwidth consumed), per-IP (pay for each unique IP allocated), and unlimited/flat-rate (fixed monthly fee for unlimited bandwidth). Each optimizes for a different customer profile.
Per-GB is the most common for residential proxies because bandwidth correlates with value delivered. Per-IP dominates datacenter and ISP proxies where you need dedicated, static addresses. Unlimited plans exist at the budget end but typically come with severe quality or speed limitations.
Per-GB pricing charges for the data transferred through the proxy. At KnoxProxy, only successful (2xx) responses are billed. This is critical: a provider that bills on all traffic, including failures, has a hidden markup proportional to your failure rate.
The effective price formula: (list price per GB) / (success rate) = effective cost. At $2.10/GB with 99.2% success, the effective cost is $2.12/GB. At $5.88/GB with 99.5% success (but billing on failures), effective cost is $5.91/GB plus the wasted bandwidth from failed attempts.
The metric that matters is cost per successful request on your targets -- not the rate card price.
Per-IP pricing makes sense when you need a static IP that does not change. ISP proxies and dedicated datacenter proxies are typically sold this way. You pay per IP per month, and bandwidth is either unlimited or generously capped.
The trade-off: per-IP is predictable for budgeting but wasteful if you do not fully utilize each IP. Calculate your expected requests per IP per day to determine if per-IP or per-GB is more economical.
Regardless of the model, the metric that matters is cost per successful request on your targets. Collect these three numbers during a trial: average page size in bytes, success rate on your actual targets, and the provider's billing rules (2xx-only vs all traffic). Multiply them out to get your cost per successful page load.
Run this calculation against at least two providers before committing. A 30% price difference on the rate card can become a 50% difference in effective cost once billing rules and success rates are factored in.
Do not compare sticker prices. Calculate cost per successful request on your targets using a free trial. The billing rules -- especially whether failed requests are charged -- often matter more than the listed per-GB rate.
Start free, run a thousand requests, and log the exit IPs. The difference is measurable in the first batch.