Skip to the answer
Proxy Costsobserved 2026-07-29

Answer

A question answered from the dataset, with the figure, the date it was observed and the source it came from.

How much do proxies cost per GB in 2026?

For conflict-free residential plans with an exact published checkout, the observed price per GB in this 2026 snapshot runs from $2 to $10 at a 1 GB tier and from $0.4999 to $5.40 at a 100 GB tier. The 1 GB endpoints are ThorData and ProxyScrape; the 100 GB endpoints are Evomi Core and Infatica IPv6. Prices were observed on July 25, 2026. ThorData residential pricing, retrieved 2026-07-25T17:34:28Z · ProxyScrape residential pricing API, retrieved 2026-07-25T16:33:08.221Z · Evomi residential pricing, retrieved 2026-07-25T16:33:08.221Z · Infatica residential pricing, retrieved 2026-07-25T15:57:28Z

There is no single 2026 proxy price per GB. The answer changes with proxy type, traffic quantity, checkout structure, and the time available to use the traffic. A rate taken from a large tier cannot be applied to a smaller purchase unless that smaller checkout is actually published.

The cleanest buyer answer is therefore a set of observed ranges at exact quantities. This page uses residential plans because traffic is their dominant public billing unit. Datacenter and ISP products are often sold per IP, so forcing them into a per-GB table would manufacture a unit the vendor does not sell.

Residential price per GB at four exact quantities

The table includes only rows that meet three conditions: the row is residential, its source state is not conflicting, and its tier contains both the exact traffic quantity and an exact published checkout total. It excludes quote-only plans, configurators, larger tiers that merely cover the request, and rows that publish a rate without a checkout total.

Exact tier Qualifying rows Observed price per GB range Median price per GB Endpoint evidence
1 GB 15 $2 to $10 $5.50 ThorData residential pricing, retrieved 2026-07-25T17:34:28Z · ProxyScrape residential pricing API, retrieved 2026-07-25T16:33:08.221Z
10 GB 10 $0.792 to $5 $2.95 GeoNode residential pricing, retrieved 2026-07-25T15:42:45Z · Shifter residential documentation, retrieved 2026-07-25T15:34:20Z
50 GB 9 $0.693 to $5 $2.45 GeoNode residential pricing, retrieved 2026-07-25T15:42:45Z · Rayobyte proxy pricing, retrieved 2026-07-25T15:53:44Z
100 GB 16 $0.4999 to $5.40 $2.425 Evomi residential pricing, retrieved 2026-07-25T16:33:08.221Z · Infatica residential pricing, retrieved 2026-07-25T15:57:28Z

The medians are computed from the exact tier totals in the complete observed plan table, not copied from vendor marketing. At an even-sized checkpoint, the median is the midpoint of the two central rates after sorting.

The cohort changes between quantities. The lower 100 GB endpoint is not the 1 GB winner extended at a guessed rate. It is evomi-core-residential-monthly, whose recorded monthly tier includes 100 GB. The upper endpoint is infatica-residential-ipv6-monthly, also at an exact 100 GB tier. Evomi residential pricing, retrieved 2026-07-25T16:33:08.221Z · Infatica residential pricing, retrieved 2026-07-25T15:57:28Z

Checkout total and unit rate answer different questions

A buyer does not pay a unit rate in isolation. GeoNode’s exact 10 GB row pairs a $7.92 checkout with a $0.792 per-GB rate. Its exact 50 GB row pairs a $34.65 checkout with a $0.693 rate. The lower unit price requires the larger receipt. GeoNode residential pricing, retrieved 2026-07-25T15:42:45Z

That distinction becomes larger at the high-volume end. GeoNode records a $0.27 per-GB rate at 100,000 GB, attached to a $27,000 checkout. Applying that rate to a 10 GB purchase would imply a checkout the vendor does not publish. GeoNode residential pricing, retrieved 2026-07-25T15:42:45Z

The effective rate can also depend on unused traffic. If a buyer needs 50 GB and purchases a 100 GB package, dividing the receipt by the purchased amount describes the tier’s sticker rate. Dividing the same receipt by the 50 GB requirement describes the buyer’s effective cost. Both calculations can be correct, but they answer different questions and must state their denominator.

Expiry changes what a low rate buys

The 100 GB endpoint from Evomi has a 30-day traffic window and no rollover on the provider’s page in the recorded fields. GeoNode’s traffic is explicitly non-expiring. Those terms are real product differences, so the lowest rate should not be detached from the usage schedule. Evomi bandwidth terms, retrieved 2026-07-25T16:33:08.221Z · GeoNode bandwidth terms, retrieved 2026-07-25T15:42:45Z

If expiry is null in the dataset, the term is not published. It is not treated as non-expiring. The same rule applies to rollover, concurrency, and every other decision field. A low rate with an unpublished traffic lifetime is less calculable than a slightly higher rate with a clear lifetime.

Other proxy types need their own unit

The dataset has exact traffic-priced observations outside residential, but coverage is thinner. At 50 GB, the exact conflict-free datacenter checkout from infatica-shared-datacenter-monthly is $29, or $0.58 per GB. The exact mobile observations run from proxyrack-mobile-monthly at $50, or $1 per GB, to froxy-mobile-50 at $300, or $6 per GB. Infatica datacenter pricing, retrieved 2026-07-25T15:57:28Z · ProxyRack mobile pricing, retrieved 2026-07-25T16:03:55Z · Froxy mobile pricing, retrieved 2026-07-25T17:34:28Z

Those are observations, not broad market ranges. Only one datacenter row supplies an exact published 50 GB total under this strict filter, while three mobile rows do. Per-IP datacenter or ISP plans remain outside the calculation because their traffic allowance is either a different field or not published.

The practical 2026 answer

Start with the proxy type and exact monthly traffic requirement. Select only purchasable tiers that cover that amount. Keep the whole checkout, the included GB, expiry, rollover, and source state beside the computed rate. If a plan is sold per IP, by concurrency, with unrestricted transfer, through a configurator, or by quote, leave its per-GB value null.

For exact residential checkouts in this snapshot, the observed buyer-sized ranges are the four ranges above. They are dated July 25, 2026, and should be re-observed before payment. A different traffic quantity changes both the plan cohort and the rate.