How much do residential proxies cost per GB in 2026?
Residential proxy cost per GB depends on billed usage. Compare allowances, retries and usable results to calculate bandwidth needs before choosing a plan.
Residential proxy cost per GB in 2026 is the amount billed for bandwidth divided by the gigabytes that charge covers; compare it against the gigabytes you actually use, not just the allowance you buy. The advertised rate does not describe your cost per usable result: unused allowances, retries and browser resources change that calculation.
TL;DR
- Residential proxy cost per GB measures bandwidth, not successful data collection.
- Node4 residential proxies serve developers and businesses needing residential exits across 170+ countries.
- Compare billed traffic, allowance expiration and usable results before choosing a plan.
- Measure HTTP requests and browser automation separately because their resource loads differ.
Why this matters
Buy against your workload, not a headline bandwidth rate. A data team needs valid records. An agency needs completed client jobs. A developer needs requests that return the expected content without exhausting the traffic allowance.
Those outcomes are different from transferring bytes. A response containing a block page still transfers data, while a browser can fetch resources your parser never uses. Check the billing rules before treating every downloaded byte as useful work.
For a technical starting point, use the curl authenticated proxy workflow. Establish a working request before you compare commercial terms. The residential proxies page lists current plans and country coverage.
How much do residential proxies cost per GB in 2026?
Calculate residential proxy cost per GB from the current plan terms. Keep the advertised bandwidth rate separate from the effective rate of your actual usage, then connect both to the output your job produces.
Use these calculations:
Advertised allowance rate = bandwidth charge / included GB
Effective usage rate = bandwidth charge / actual billed GB used
Cost per usable result = total job charge / accepted resultsThe first calculation describes the offer. The second describes how much value you get from the allowance. The third describes the business outcome. None replaces the others.
For a usage-billed arrangement, divide the charge attributable to traffic by the corresponding billed traffic. For an allowance-based arrangement, also calculate the rate against the portion you consume before expiration. Keep any separate fees visible rather than quietly folding them into an unexplained denominator.
Compare identical units and identical billing periods. A monthly allowance and a balance with different expiration terms are not equivalent just because both display a per-GB rate.
Check what the bandwidth unit means
A gigabyte and a gibibyte are different units. Under the SI decimal convention, 1 GB is 1,000,000,000 bytes. Under the IEC binary convention, 1 GiB is 1,073,741,824 bytes, or 1,024 MiB.
| Unit | Bytes | Best for | Limitation |
|---|---|---|---|
| GB | 1,000,000,000 | Comparing terms explicitly stated in decimal gigabytes | Do not treat it as interchangeable with GiB |
| GiB | 1,073,741,824 | Interpreting measurements explicitly stated in binary gibibytes | Convert it before comparing with a decimal GB allowance |
These are unit definitions, not proxy performance figures. Check the provider's billing definition and the unit reported by your monitoring system. A label alone does not establish that both systems use the same denominator.
Compare billing arrangements without mixing their denominators
Residential proxy billing terms determine which calculation matters most. Evaluate the arrangement actually offered, rather than assuming that every provider uses the same model.
| Arrangement | Best for | What to verify | Main limitation |
|---|---|---|---|
| Usage billing | Jobs with uncertain traffic requirements | Which bytes count and whether minimum commitments apply | Traffic consumption needs active monitoring |
| Prepaid traffic allowance | Workloads with a measured traffic requirement | Expiration, unused balance and additional usage terms | Unconsumed traffic changes the effective usage rate |
| Recurring traffic allowance | Repeated jobs with a predictable schedule | Renewal, rollover and behavior at the allowance limit | The billing schedule must match the workload |
This table compares billing arrangements, not named provider plans. A subscription can include a prepaid allowance, and a usage arrangement can include a commitment. Read the actual terms rather than assigning a plan to a category from its label.
Choose usage billing for uncertain demand only after checking commitments. Choose an allowance only after measuring consumption. The advantage of either arrangement disappears if its conditions do not fit your job schedule.
For an agency, allocate consumption by client or project. A shared account total tells you what the account consumed, but it does not tell you which contract generated that consumption. Tag jobs before they run so you can reconcile usage afterward.
Measure a representative job before buying bandwidth
A useful residential proxy budget starts with a representative workload. Keep the target, extraction logic and validation rules consistent during the measurement. Otherwise, you are comparing different jobs rather than different proxy arrangements.
- Define an accepted result. Specify the fields, page state or response structure your job needs. An HTTP success response is not enough if the required data is absent.
- Select representative targets. Include the page types and locations your production job actually uses. Do not measure a lightweight endpoint and extrapolate it to a browser-heavy workflow.
- Record application traffic. Capture request counts, response sizes, redirects and retries. Separate successful extraction from merely completed requests.
- Record provider usage. Capture the account's billed traffic over the same interval. Isolate the test from unrelated jobs where possible.
- Reconcile the difference. Investigate discrepancies before extrapolating. Your application's counters and the provider's meter describe different observation points.
- Calculate output efficiency. Divide measured billed bytes by accepted results. Use that relationship to estimate the bandwidth requirement for the production workload.
The useful comparison remains the same: hold the task constant and measure both bandwidth and accepted output. Changing countries, rendering modes or extraction requirements during the comparison makes the result harder to interpret.
Inspect a request with curl
The following shell commands expect you to supply your target URL and authenticated proxy URL. Keep credentials out of source control and avoid publishing terminal output that exposes them.
: "${TARGET_URL:?Set TARGET_URL}"
: "${PROXY_URL:?Set PROXY_URL}"
curl \
--proxy "$PROXY_URL" \
--location \
--silent \
--show-error \
--output response.body \
--write-out 'status=%{http_code}\nbody_bytes=%{size_download}\nredirects=%{num_redirects}\n' \
"$TARGET_URL"The command follows redirects, saves the response body and reports curl's request measurements. Inspect the saved body to verify that it contains the expected content rather than a login page or challenge.
Curl's download measurement is not proof of the provider's billed traffic. Treat it as an application-side observation. Redirects, request traffic and billing rules require separate attention when you reconcile the account meter.
An HTTP request also does not reproduce a browser session. It does not automatically load a page's scripts, images and other resources. Measure the execution mode you intend to deploy.
Why measured bandwidth varies
The workload determines how many bytes you transfer to obtain an accepted result. These factors explain differences without assuming any provider's performance or billing policy:
- Response payload: HTML, structured responses and downloadable files have different byte sizes. Measure the content your task requires.
- Browser resources: Browser automation can request scripts, stylesheets, images and fonts beyond the initial document. Those requests belong in the workload measurement.
- Retries: A repeated attempt sends another request and can receive another response. Record retries separately so failed work does not disappear inside a total.
- Redirects: Following a redirect creates another request. Inspect the chain when traffic differs from your initial estimate.
- Compression: Content encoding changes transferred payload size. Check whether application measurements describe transferred or decoded content before comparing them.
Do not assign a fixed multiplier to these factors without measuring your job. Their impact depends on what your application fetches and how it processes the response.
Concurrency is a separate control. It determines how much work runs at once, not how many bytes each response contains. Read the aiohttp high-concurrency scraping workflow when tuning asynchronous requests, and keep bandwidth accounting separate from connection limits.
Calculate a bandwidth requirement from your own measurements
Use measured billed bytes and accepted results to estimate the traffic requirement. The script below accepts your measurements without inserting example usage figures or assumed success rates.
from decimal import Decimal
billed_bytes = Decimal(input("Measured billed bytes: "))
accepted_results = Decimal(input("Accepted results: "))
planned_results = Decimal(input("Planned results: "))
if billed_bytes <= 0:
raise ValueError("Measured billed bytes must be positive")
if accepted_results <= 0:
raise ValueError("Accepted results must be positive")
if planned_results <= 0:
raise ValueError("Planned results must be positive")
bytes_per_result = billed_bytes / accepted_results
planned_bytes = bytes_per_result * planned_results
planned_gb = planned_bytes / Decimal("1000000000")
print(f"Measured billed bytes per result: {bytes_per_result}")
print(f"Estimated decimal GB required: {planned_gb}")This is a linear estimate, not a promise about future consumption. It assumes the measured mix of payloads, retries and accepted results remains representative of the planned run.
For a production budget, retain the measurements behind the estimate. Record the target set, execution mode, location selection and measurement interval. When the job changes, rerun the measurement rather than preserving an outdated bandwidth assumption.
Keep estimates for materially different workflows separate. Combining browser rendering and direct HTTP extraction into a single average can conceal which task consumes the allowance. Aggregate them only after calculating each workload's requirement.
What should you check before choosing a residential plan?
Start with billing definitions, then verify that the service fits the job. Ask for clear answers on the following points:
- Which traffic directions and request outcomes count toward usage?
- Does the allowance expire, renew or roll over?
- What happens when the allowance is exhausted?
- Does location targeting change the applicable terms?
- Which protocols and authentication methods does your client need?
- What are the session rules for a workflow that needs a stable exit?
Treat unanswered questions as unresolved purchase criteria. Do not infer an expiration policy, authentication method or session duration from another product sold by the same provider.
Node4 residential proxies are best for developers and businesses that need residential exits across 170+ countries. Those residential addresses come from an upstream supplier and are resold, not owned by Node4. Country coverage is relevant to location-specific collection; it is not evidence that a particular target accepts an exit or that the workload will meet your bandwidth estimate.
Validate the required location and target before committing a production workload. A country selection, a successful connection and an accepted data result are separate checks.
Is a lower per-GB rate always better?
A lower per-GB rate is better only when the billing terms and workload outcome are comparable. If you buy unused traffic or transfer more bytes to obtain the same accepted output, the lower advertised rate does not settle the decision.
Compare accepted results over the same measurement interval. Keep failed attempts and unused allowances visible instead of reporting only the successful request subset.
Do failed requests count toward residential proxy usage?
A failed request can transfer bytes, but whether those bytes are billed depends on the provider's terms. Distinguish connection failure, an error response and a completed response that fails your content validation.
Record those categories separately. They describe different technical failures and require different fixes.
Is bandwidth the same as concurrent connections?
Bandwidth measures transferred data; concurrent connections measure simultaneous connections. A connection allowance does not state how many gigabytes your workflow consumes.
For a buying decision, compare each limit against the matching measurement. Do not convert a connection count into a traffic allowance.
FAQ
How do I calculate residential proxy cost per GB?
Divide the bandwidth charge by the gigabytes covered by that charge. For an allowance-based plan, also divide the charge by the traffic you actually consume to calculate your effective usage rate.
What is the best residential proxy billing model for an irregular job?
A usage-billed arrangement fits irregular demand when its commitment and billing rules match the job. Check minimum commitments and measure the workload before choosing it.
Does a residential proxy bill include blocked responses?
A blocked response transfers data, but whether it counts toward billed usage depends on the provider's terms. Track blocked responses separately from accepted results.
Is 1 GB the same as 1 GiB?
1 GB is 1,000,000,000 bytes; 1 GiB is 1,073,741,824 bytes. Convert both measurements to the same unit before comparing a traffic allowance with your application's usage.
Can I use curl download size as my proxy bill?
Curl download size is an application-side measurement, not a billing statement. Reconcile it with the provider's usage meter and documented traffic rules.
Why does browser automation use more than the initial HTML response?
Browser automation can fetch scripts, stylesheets, images and other resources after loading the initial document. Measure the complete browser workflow rather than only the initial response.
How should an agency allocate residential proxy bandwidth?
An agency should track usage by client or project and reconcile those measurements with account-level billed traffic. Keep shared overhead and failed work visible instead of assigning everything to accepted results.
One last thing
Define success before calculating cost. A completed transfer is not necessarily usable data. If your validation rejects the content, that transfer belongs in the workload's consumption but not in its accepted-result count.
Keep that distinction in your reporting. It turns a bandwidth comparison into a decision about the output your team actually needs.