Unlimited bandwidth proxies promise a flat bill no matter how much traffic your scrapers, bots or monitoring jobs push through them — and that promise is exactly why “unlimited” is the most abused word in the proxy industry. Every plan runs on hardware with real costs: uplinks, rack space, and the residential or ISP connections your traffic finally exits through. A vendor selling truly unconstrained transfer at a flat price is either subsidizing you, slowing you down, or betting you will never read the fine print.
That does not make unlimited plans a bad deal. It makes them a pricing envelope rather than a physics guarantee. Some vendors mean it honestly and cap only your port speed; others let you run full speed for two weeks, then quietly throttle you to a trickle once you cross a monthly threshold buried in the acceptable-use policy. The gap between those two outcomes is worth hundreds of dollars a month, and this guide gives you the vocabulary, the math and the verification steps to tell them apart before you commit.
The four pricing models behind unlimited bandwidth proxies
When a plan page says “unlimited,” it is describing one of four very different billing models. Identifying which one you are looking at answers most of your questions before you ask them.
- Per-GB: you pay for the traffic you actually use. Costs are predictable and audits are easy, but heavy scraping gets expensive fast, and every retry burns money twice.
- Unlimited with fair-use: a flat monthly rate with a soft cap hidden in the acceptable-use policy. Cross the threshold and the provider throttles your speed, deprioritizes your traffic or nudges you to upgrade. This is the most common “unlimited” offer on rotating residential pools.
- Per-port (per dedicated IP): you pay per port or per static IP with pooled or generous bandwidth. Ideal when you need many long-lived identities doing light-to-medium work, and it pairs naturally with static ISP pool sizing.
- True unlimited with capped speed: genuinely no traffic meter, but the port speed is fixed — say 50 or 100 Mbps. Rare, usually on ISP or static residential plans, and often the cheapest predictable option for round-the-clock automation.
Checklist: estimate your real bandwidth needs
Most buyers overestimate by 5–10× because they guess instead of measuring. Run this six-step estimate before you compare any prices.
- Measure the payload of one representative page or API response — save the raw bytes, not the rendered page. A product page without images is often 60–120 KB.
- Multiply the payload by requests per hour for a single worker.
- Multiply by the number of parallel workers and the active hours per day.
- Multiply by 30 days for a monthly baseline, then add 30–50% overhead for retries, duplicate fetches, images, fonts and challenge pages.
- Model your peak hour, not your average — fair-use throttles and rate limits bite at peaks first.
- After the first week, compare your projection against the provider’s dashboard counters and recalibrate.
Worked example: a product page at 80 KB × 10,000 requests per day ≈ 0.8 GB per day ≈ 24 GB per month. Add 40% overhead and your budget number is roughly 34 GB — a figure that instantly tells you whether per-GB or flat pricing wins for your workload.
Comparison: the four models side by side
| Pricing model | How billing works | Best suited for |
|---|---|---|
| Per-GB | Metered transfer, prepaid or postpaid | Light, spiky or irregular workloads; clean budgeting |
| Unlimited (fair-use) | Flat rate; throttle or deprioritization after a threshold | Medium workloads that tolerate variable speed |
| Per-port / per-IP | Fixed price per dedicated port or static IP | Many long-lived sessions, accounts and checkout flows |
| True unlimited (capped speed) | No traffic meter, fixed port speed | 24/7 automation with steady, predictable demand |
How to choose and verify an unlimited plan
- Write down your monthly GB estimate from the checklist above — one number, in writing.
- Map it to a model: under roughly 50 GB per month, per-GB is usually cheapest; above that, flat models start winning.
- Read the acceptable-use policy end to end and search for the words “throttle,” “fair,” “deprioritize” and “excessive.”
- Ask support directly: what happens at the threshold, and is the cap measured per day, per week or per month?
- Run a sustained download through the proxy at hour one, hour 24 and hour 168, and compare throughput — a proper how to test a proxy routine catches slow-burn throttles that a single speed test will miss.
- Recheck after the first billing cycle and keep your own traffic counters; never rely on the dashboard alone.
Common mistakes
- Buying “unlimited” for heavy scraping without reading the AUP. The throttle clause is always there; the only question is whether it sits above or below your real usage.
- Estimating with average load instead of peak. A job that averages 2 GB/hour but peaks at 20 GB/hour will trip fair-use rules written around sustained averages.
- Forgetting overhead. Retries, duplicate content, images and challenge pages routinely add 30–50%; plans bought at exactly 1.0× estimated usage fail within weeks.
- Overbuying dedicated capacity. Paying for twenty static ports when rotating residential per-GB would cost a third as much is the mirror-image error — match the model to the workload, not to the brochure.
FAQ
Is unlimited bandwidth on proxies ever real?
Yes, but almost always with a speed cap instead of a traffic cap. True unlimited plans exist on ISP and static residential products where you buy a port at a fixed speed; because the speed is capped, the maximum monthly transfer is mathematically bounded and the vendor can price it honestly. On rotating residential pools, “unlimited” almost always means fair-use.
What hidden throttles should I look for?
Four patterns: a monthly threshold after which speed drops (often 3–10× slower), deprioritization during peak hours, per-day burst caps that reset at midnight UTC, and concurrent-connection limits that cap throughput indirectly. All four live in the AUP or the plan’s fine print — none appear on the pricing page.
How do I estimate how many GB I need?
Payload × requests per hour × active hours × days, plus 30–50% overhead. Measure the payload from a real response, not a guess. The worked example above — 80 KB × 10,000 requests/day ≈ 24 GB/month before overhead — takes five minutes and converts a marketing label into a number you can budget against.
Per-GB or flat rate for automation?
Per-GB wins for light, spiky or experimental workloads because idle time costs nothing. Flat rate wins when utilization is high and steady: if your projected monthly GB costs more at per-GB rates than the flat plan, switch. For 24/7 automation on dedicated IPs, per-port true-unlimited plans are usually the cheapest predictable option — just verify the speed cap covers your peak hour.
Unlimited bandwidth is real — as a speed-capped port, a fair-use envelope or an honestly metered bill. The buyers who win are the ones who do the five-minute math, read the AUP and measure their own traffic instead of trusting a dashboard. FTProxy residential proxies publishes exactly what every plan includes — per-GB, per-port and capped-speed options with no hidden throttles — so your bill never surprises you.