Per-GB pricing means you pay for the traffic that actually passes through the proxy, usually on top of a big rotating pool. To see how this model compares to per-IP, per-port, and pool subscriptions, see proxy pricing models overview.
How Per-GB Proxy Billing Works
You are charged for gigabytes transferred through the proxy gateway, typically counting both download and upload. In most cases 1 GB is treated as 1024 MB, though you should confirm the vendor’s definition.
Under Per-GB plans you authenticate to a gateway that fronts a large IP pool. You are not buying specific IPs, you are buying the right to move traffic through the pool. This model is common on residential and mobile networks where continuous rotation is expected. For a quick refresher on networks and pools, see our overview of proxy network types.
What counts as traffic
Everything that traverses the proxy usually counts: request and response bodies, headers, TLS handshakes, and data from retries. Some providers exclude certain control traffic, but most simply meter bytes in both directions. If your workflow pulls images, video, or large JSON payloads, your GB spend rises fast.
Measuring a gigabyte
Vendors often use 1 GB = 1024 MB. A few round differently. Always check the fine print in dashboards or docs, then baseline your average MB per request with a small pilot before buying a large bundle.
What Actually Consumes Your GBs
All bytes over the tunnel are billable unless a vendor states otherwise. Failures and retries still burn traffic because the proxy carried those bytes.
Small requests add up. TLS adds overhead at connection start, headers add overhead on every request, and redirects can double or triple transfer. If your crawler fetches full image assets or client-side bundles by default, you will pay for them.
Rule of thumb. First measure the average MB per successful request on the real target, then multiply by your monthly request count and add a safety factor for retries.
Why Choose Per-GB
Per-GB is ideal when you need very large IP diversity and frequent rotation, or when usage is uncertain and you want costs to scale with activity. You are paying for outcomes in bytes rather than for a fixed number of IPs that may sit idle.
This model aligns with short projects, proofs of concept, and seasonal traffic. It also scales to very high volumes because vendors often offer tiered discounts as your committed GB rises.
Bundles, Pay-Go, Top-Ups, and Rollover
Per-GB comes in two main billing patterns. Prepaid bundles give you a fixed quota that stops service at the cap unless you upgrade or top up. Metered pay-as-you-go bills your actual usage, often with optional auto-recharge.
Rollover varies. Monthly bundles usually reset at renewal and unused GBs expire. Wallet-style credit often persists until consumed. Subscriptions auto-renew; true pay-go relies on manual or auto top-ups.
Enable hard guards where possible. Set “stop at quota,” spend caps, and threshold alerts to avoid runaway jobs. For the policy details that can affect throughput or session behavior, review your provider’s usage limits.
Pros and Cons
Per-GB gives flexible access to huge rotating pools and maps spend to actual use, but costs can spike on bandwidth-heavy targets and budgeting can be tricky without safeguards.
Pros
- Access to very large, rotating pools without buying IPs.
- Costs track real work, good for intermittent or variable workloads.
- Volume tiers can drop your effective $ per GB at scale.
Cons
- Media-heavy pages and retries consume GB quickly.
- Some vendors bill failures, redirects, and handshake overhead.
- Budgeting is harder if request sizes vary across targets.
Hidden Costs To Watch
Overhead, rounding, and overage tiers often matter more than the sticker price. Read how the meter increments and where the plan breaks change.
- Protocol overhead: TLS handshakes and headers add non-payload bytes. Many tiny requests can burn more than a few large ones.
- Minimum increments and rounding: Per-request rounding to whole MBs will inflate totals for small calls.
- Overage pricing and forced upgrades: Exceeding a bundle may trigger higher per-GB rates or require a plan jump.
- Taxes and payment frictions: VAT, currency conversion, and processor fees change the real $ per GB you pay.
What $/GB Looks Like In Practice
Budget P2P-style pools cluster around about $1 per GB. Typical retail sits roughly $3 to $15 per GB. Large enterprise commits can land around $0.5 to $2 per GB.
These are directional patterns intended for planning and vendor comparisons.
| Network or plan context | Typical retail range | Notes |
| Budget P2P-style residential | ~$1/GB | Few guarantees, large pools, basic support |
| Standard retail residential/mobile | ~$3–$15/GB | Rotation included, dashboards and APIs, tiered discounts |
| High-volume enterprise commits | ~$0.5–$2/GB | Contracted volumes, negotiated SLAs |
| Rotating datacenter by GB | Often below retail Res/Mob | Cheaper per byte, lower diversity than household IPs |
Ranges vary by geo coverage, tooling, support, and contractual terms.
Compare Providers Beyond Sticker Price
Do not judge providers by posted $/GB alone. Measure success rate, average MB per page, retry behavior, and how smoothly rotation works on your targets.
Run a pilot on the real sites you care about. Record p50 and p95 bytes per successful request and the retry rate. A cheaper $/GB with poor success can burn more total GB than a slightly pricier service that succeeds on the first attempt. Also consider concurrency policies and whether access is private or shared on key gateways, since sharing can affect stability under load. If you need a refresher on access semantics, see private vs shared proxies.
Quick sizing example. If your measured average is 0.35 MB per successful request and your pipeline makes 5 million successful requests per month, baseline consumption is roughly 1,750 MB, or about 1.71 GB. Add overhead for retries and redirects, then round up to the next plan breakpoint to avoid overage.
Control Your Spend
In two lines: Use the controls your vendor offers, then reduce payload. Alerts, hard caps, and simple content hygiene lower your $/GB without changing targets.
Practical steps:
- Set guards: hard quota, spend caps, webhook or email alerts at 50, 75, and 90 percent of plan.
- Trim payload: avoid downloading images, video, or oversized bundles by default. Prefer HEAD where allowed and use partial content or range requests when feasible.
- Reduce waste: cache stable assets, deduplicate URLs, and fix retry loops that follow redirect chains.
- Connection hygiene: reuse sessions when targets allow it to amortize TLS handshakes.
- Buy at breakpoints: a bigger tier can be cheaper than paying overage on a smaller bundle.
Best-Fit Use Cases, Pitfalls, and Fixes
Per-GB fits heavy-rotation tasks, proofs of concept, and seasonal or uncertain workloads. Most failures come from loops, asset bloat, or undersized plans.
Well-suited
- High-diversity workflows where identity can rotate frequently.
- Short projects or pilots where you want to cap spend exactly.
- Work with seasonality or traffic spikes.
Common pitfalls and fixes
- Hidden asset load: Your crawler fetches images or video. Fix by blocking static assets or using text-only endpoints.
- Retry storms: Script retries on 429/5xx without backoff. Fix by adding jittered backoff and circuit breakers.
- Undersized plan: You buy 5 GB but consistently need 8–10 GB. Fix by stepping up one tier or enabling friendly auto-upgrades with alerts.
- Per-request rounding: Tiny calls billed as full MBs. Fix by batching or using endpoints that return more per call.
FAQ
Does upload count, or only download?
Most vendors count both directions because all bytes traverse the proxy. Assume upload is billable unless the provider explicitly excludes it.
Is 1 GB always 1024 MB?
Often yes, but not always. Check the vendor’s documentation or dashboard. Use their definition for sizing and alerts.
Can I roll unused GBs into the next month?
Monthly bundles usually reset and unused GBs expire. Wallet-style credit in true pay-go often persists until consumed.
Why is my bill higher than the bytes I expected from payload sizes?
Overhead from TLS, headers, redirects, retries, and per-request rounding can explain the gap. Measure end-to-end bytes on your actual jobs.
How do I stop runaway spend?
Enable “stop at quota,” set spend caps and alerts, and add backoff with retry limits in your code. Review usage limits to understand provider-side controls.