---
title: "VPN Speed Test Results 2026 — Our Benchmark Data | vpnrank.io"
description: "Full VPN speed benchmark data: 8 VPNs measured on a 1 Gbps US East reference line, plus regional results from Tokyo, London, Dubai, São Paulo, Shanghai, and 10 more locations. Updated quarterly."
url: https://vpnrank.io/vpn-speed-test
language: en
updated: 2026-10-07
---

# VPN Speed Test Results

ExpressVPN was the fastest VPN in our Q3 2026 benchmark: 478 Mbps on a 1 Gbps wired fiber line in US East, on a nearby server with Lightway-UDP, tested 25 August 2026. IPVanish (468 Mbps) and NordVPN (462 Mbps) finished within 4% of it; the slowest of the 8 VPNs we benchmark, TotalVPN on OpenVPN-UDP, reached 415 Mbps, 13% behind.

Across 15 regional measurements, NordVPN was fastest in 8 and ExpressVPN was fastest in 7; ExpressVPN led on all 4 routes through national filtering. Every figure is computed from the dataset below, measured per [our methodology](https://vpnrank.io/how-we-test) and refreshed quarterly.

Last tested 25 August 2026 · Page updated 7 October 2026

## Does a VPN slow down your internet?

Yes, and how much depends on your line. In our Q3 2026 benchmark (25 August 2026), on a 1 Gbps fiber line the 8 VPNs we benchmark delivered 415–478 Mbps (42–48% of the line, where the VPN, not the line, becomes the ceiling); on 200–300 Mbps home lines with a nearby server, the providers we measured there (NordVPN and ExpressVPN) kept 80–93%; long-haul routes and routes through national filtering using obfuscation kept 55–65%. For streaming the loss rarely matters: even the slowest result in the dataset, 28 Mbps (NordVPN, Tehran), is above the 15 Mbps Netflix recommends for 4K.

Share of line speed kept with a VPN, by line speed and route, Q3 2026 benchmark
| Route | Line speed | Speed kept | Measured from |
| --- | --- | --- | --- |
| Reference line, nearby server | 1000 Mbps | 42–48% (all 8 VPNs) | US East (wired fiber) |
| Nearby server | 1000 Mbps | 40–46% | Tokyo, London |
| Nearby server | 500 Mbps | 74–83% | Madrid, Milan |
| Nearby server | 300 Mbps | 80–88% | Mexico City, São Paulo |
| Nearby server | 200 Mbps | 81–93% | Jakarta, Istanbul, Buenos Aires |
| Distant exit, no obfuscation | 200 Mbps | 79–83% | Cairo |
| Long-haul exit (other continent) | 500 Mbps | 56–59% | Bangkok |
| Through national filtering (obfuscated) | 50–500 Mbps | 55–65% | Dubai, Riyadh, Shanghai, Tehran |

"Speed kept" is a VPN's download divided by the line's speed without a VPN. Each range spans every provider measured on that row: all 8 on the reference line; ExpressVPN in all 15, NordVPN in all 15, Surfshark in 6 of the 15 regional locations. Per-provider figures are in the tables below.

## Reference Benchmark — US East, 1 Gbps

Wired fiber connection, nearby server, each provider's recommended protocol. Average of multiple runs at different times of day. Q3 2026, tested 25 August 2026.

| # | VPN | Protocol | Download (Mbps) | Speed retention | Behind fastest |
| --- | --- | --- | --- | --- | --- |
| 1 | ExpressVPN | Lightway-UDP | 478 | 48% | — |
| 2 | IPVanish | WireGuard | 468 | 47% | 10 Mbps (2.1%) |
| 3 | NordVPN | NordLynx | 462 | 46% | 16 Mbps (3.3%) |
| 4 | Proton VPN | WireGuard | 445 | 45% | 33 Mbps (6.9%) |
| 5 | CyberGhost | WireGuard | 438 | 44% | 40 Mbps (8.4%) |
| 6 | Private Internet Access | WireGuard | 430 | 43% | 48 Mbps (10.0%) |
| 7 | Surfshark | WireGuard | 421 | 42% | 57 Mbps (11.9%) |
| 8 | TotalVPN | OpenVPN-UDP | 415 | 42% | 63 Mbps (13.2%) |

## How These Speed Tests Were Run

What the dataset on this page records, and what it doesn't.

Cycle

Q3 2026, last tested 25 August 2026. The dataset was first published on 11 June 2026; the changelog below lists every refresh.

Reference line

US East (wired fiber), 1000 Mbps without a VPN, connecting to a nearby US East server.

Protocol

Each provider's recommended protocol: Lightway-UDP (ExpressVPN), WireGuard (5 providers), NordLynx (NordVPN), OpenVPN-UDP (TotalVPN).

What is measured

Download throughput in Mbps. Each published figure is the average of multiple runs at different times of day, not a single best case.

Regional runs

15 locations on local lines from 50 to 1000 Mbps: ExpressVPN in all 15, NordVPN in all 15, Surfshark in 6. Routes in UAE, Saudi Arabia, China and Iran go through obfuscated protocols.

Not in this dataset

Per-run logs, run counts, app versions, upload speed and latency. The page publishes download averages only.

## Regional Measurements

In-region and long-haul results: ExpressVPN in all 15, NordVPN in all 15, Surfshark in 6 of the 15 locations. Baselines reflect local line speeds; restrictive markets (UAE, Saudi Arabia, China and Iran) are measured through obfuscated protocols.

| Location | Baseline (Mbps) | Results (Mbps) | Fastest | Notes |
| --- | --- | --- | --- | --- |
| Tokyo, Japan | 1000 | ExpressVPN 455 · NordVPN 447 · Surfshark 396 | ExpressVPN (46%) | Local fiber (Nuro Hikari), Tokyo servers |
| London, UK | 1000 | ExpressVPN 441 · NordVPN 433 · Surfshark 405 | ExpressVPN (44%) | London servers from a UK fiber line |
| Madrid, Spain | 500 | NordVPN 415 · ExpressVPN 395 · Surfshark 372 | NordVPN (83%) | Madrid servers from Spanish fiber |
| Milan, Italy | 500 | NordVPN 410 · ExpressVPN 405 · Surfshark 380 | NordVPN (82%) | Milan servers from Italian fiber |
| Dubai, UAE | 500 | ExpressVPN 312 · NordVPN 305 · Surfshark 280 | ExpressVPN (62%) | UK/EU exits through Etisalat DPI (obfuscated) |
| Riyadh, Saudi Arabia | 500 | ExpressVPN 305 · NordVPN 298 · Surfshark 275 | ExpressVPN (61%) | UK/EU exits through STC DPI (obfuscated) |
| Mexico City, Mexico | 300 | NordVPN 265 · ExpressVPN 252 | NordVPN (88%) | Mexico City servers from local fiber |
| São Paulo, Brazil | 300 | NordVPN 252 · ExpressVPN 240 | NordVPN (84%) | São Paulo servers from local fiber |
| Bangkok, Thailand | 500 | NordVPN 295 · ExpressVPN 280 | NordVPN (59%) | Long-haul to London / New York exits |
| Jakarta, Indonesia | 200 | NordVPN 185 · ExpressVPN 175 | NordVPN (93%) | Jakarta servers from local fiber |
| Istanbul, Turkey | 200 | NordVPN 175 · ExpressVPN 168 | NordVPN (88%) | Istanbul servers from Turkish fiber |
| Cairo, Egypt | 200 | ExpressVPN 165 · NordVPN 158 | ExpressVPN (83%) | London exits through Egyptian ISP routing |
| Buenos Aires, Argentina | 200 | NordVPN 165 · ExpressVPN 162 | NordVPN (83%) | Buenos Aires servers from local fiber |
| Shanghai, China | 200 | ExpressVPN 130 · NordVPN 125 | ExpressVPN (65%) | Hong Kong / Tokyo exits through the Great Firewall (obfuscated) |
| Tehran, Iran | 50 | ExpressVPN 32 · NordVPN 28 | ExpressVPN (64%) | Istanbul / EU exits through national filtering (obfuscated) |

Figures are point-in-time averages, not guarantees — your speeds depend on your line, distance to the server, and local network conditions. See [How We Test](https://vpnrank.io/how-we-test) for equipment, runs, and limitations.

## VPN Speed Questions, Answered

### Which VPN is the fastest in 2026?

On our 1 Gbps reference line, ExpressVPN: 478 Mbps in the Q3 2026 cycle, 10 Mbps ahead of IPVanish. The margin at the top is small — the first three are within 4% of each other — so server choice and your own line can matter as much as the brand. Regionally the lead changes hands: NordVPN was fastest in 8 of 15 locations and ExpressVPN was fastest in 7 of 15 locations. Our [fastest VPN guide](https://vpnrank.io/best-vpn/fastest-vpn) weighs these results against price and features.

### Does the VPN protocol change the speed?

Partly. In this dataset the fastest result used Lightway-UDP, and the only provider measured on OpenVPN (TotalVPN, OpenVPN-UDP) was the slowest at 415 Mbps. Yet the 5 providers measured on plain WireGuard alone spread from 421 to 468 Mbps, so each provider's servers and network matter as much as the protocol. Our [WireGuard glossary entry](https://vpnrank.io/glossary/wireguard) explains the protocol most of them use.

### Is a VPN fast enough for 4K streaming?

Every result in this benchmark is. Netflix recommends 15 Mbps or more for Ultra HD (4K), and the slowest figure we recorded anywhere was 28 Mbps (NordVPN, Tehran). On ordinary routes the problem for streaming is rarely speed; it is whether the service blocks the VPN's address.

### Why do other VPN speed tests report a higher percentage?

Because the percentage depends on the line it's measured on. The same providers kept 80–93% of our 200–300 Mbps home lines and 42–48% of a 1 Gbps line. A test on a slower line, or on different hardware and server distance, reports a different percentage for the same VPN. Compare results only when the line speed, server distance and protocol match.

### Which VPNs does the benchmark cover?

The reference line covers 8: ExpressVPN, IPVanish, NordVPN, Proton VPN, CyberGhost, Private Internet Access, Surfshark and TotalVPN. We track prices for 9 VPNs daily; PureVPN is not part of this speed benchmark, so we quote no speed figure for it. The regional runs cover ExpressVPN in all 15, NordVPN in all 15, Surfshark in 6 locations.

### How do I test my own VPN's speed?

Run a speed test with the VPN off, then on, from the same device, and divide the second result by the first. Use a nearby server first, repeat at different times of day, and compare your percentage with the row in the table above that matches your line speed and route. A big shortfall against that row usually points at the server you picked or your Wi-Fi, not the VPN. To check that the VPN is actually carrying your traffic, use our [VPN test](https://vpnrank.io/tools/is-my-vpn-working); for the background, read [does a VPN slow down your internet](https://vpnrank.io/blog/does-a-vpn-slow-down-your-internet).

## Benchmark Changelog

-   2026-08-25 Full Q3 2026 refresh of the US East reference benchmark and 15 regional measurements.
-   2026-06-11 First public release of the benchmark dataset. Q2 2026 cycle: US East reference figures for all 8 ranked VPNs, plus 15 regional measurements for ExpressVPN, NordVPN, and Surfshark.

## Sources

Standards and vendor documentation behind the facts on this page, re-read on 7 October 2026.

-   [Netflix Help Center — Netflix-recommended internet speeds](https://help.netflix.com/en/node/306) — 15 Mbps or higher for Ultra HD (4K), 5 Mbps for 1080p, 3 Mbps for 720p.
-   [WireGuard — project homepage](https://www.wireguard.com/) — the protocol five of the eight providers were measured on.
-   [NordVPN Support — WireGuard connection on NordVPN](https://support.nordvpn.com/hc/en-us/articles/51264943286545-WireGuard-connection-on-NordVPN) — NordLynx, the default protocol in NordVPN's apps, is built around WireGuard with a double NAT system.
-   [ExpressVPN — Lightway: A VPN Protocol by ExpressVPN](https://www.expressvpn.com/lightway) — the protocol behind the fastest reference result (Lightway-UDP).

[Best Fastest VPNs](https://vpnrank.io/best-vpn/fastest-vpn) [VPN Price Index](https://vpnrank.io/vpn-price-index) [How We Test](https://vpnrank.io/how-we-test) [All VPN Guides](https://vpnrank.io/best-vpn)

Structured data (JSON-LD):

```json
{"@context":"https://schema.org","@type":"Organization","@id":"https://vpnrank.io/#organization","name":"VPNRank","alternateName":"vpnrank.io","url":"https://vpnrank.io","email":"hello@vpnrank.io","logo":{"@type":"ImageObject","url":"https://vpnrank.io/logos/favicon.png","width":500,"height":500},"description":"Independent VPN comparison. We track 9 major VPN services every day and rank them for privacy, security, speed and value.","foundingDate":"2026-03","sameAs":["https://www.youtube.com/@vpnrank","https://www.instagram.com/vpnrank.io","https://www.facebook.com/people/VPNRankio/61592338017872/"],"knowsAbout":["Virtual Private Networks","Internet Privacy","Cybersecurity","Network Encryption","VPN Protocols"]}
```

```json
{"@context":"https://schema.org","@type":"Organization","@id":"https://vpnrank.io/#editorial-team","name":"VPNRank Editorial Team","url":"https://vpnrank.io/about","parentOrganization":{"@id":"https://vpnrank.io/#organization"},"description":"A team of privacy and cybersecurity analysts who independently test VPN services for speed, security, leak protection, and value.","knowsAbout":["VPN testing methodology","WireGuard and OpenVPN protocols","DNS and IPv6 leak testing","No-logs policy audits","Encryption standards"]}
```

```json
{"@context":"https://schema.org","@type":"WebSite","@id":"https://vpnrank.io/#website","name":"VPNRank","alternateName":"vpnrank.io","url":"https://vpnrank.io","publisher":{"@id":"https://vpnrank.io/#organization"},"inLanguage":"en-US"}
```

```json
{"@context":"https://schema.org","@type":"Dataset","name":"VPNRank VPN Speed Benchmark","description":"Quarterly VPN download-speed benchmark: 8 VPN services measured on a wired 1 Gbps US East reference connection plus regional measurements from 15 worldwide locations.","url":"https://vpnrank.io/vpn-speed-test","creator":{"@id":"https://vpnrank.io/#editorial-team"},"publisher":{"@id":"https://vpnrank.io/#organization"},"dateModified":"2026-08-25","temporalCoverage":"Q3 2026","variableMeasured":"VPN download throughput (Mbps)","license":"https://vpnrank.io/terms"}
```

```json
{"@context":"https://schema.org","@type":"Article","headline":"VPN Speed Test Results 2026 — Our Benchmark Data","description":"Full VPN speed benchmark data across 8 VPNs and 15 worldwide regions, updated quarterly.","url":"https://vpnrank.io/vpn-speed-test","image":"https://vpnrank.io/opengraph-image","mainEntityOfPage":{"@type":"WebPage","@id":"https://vpnrank.io/vpn-speed-test"},"datePublished":"2026-06-11","dateModified":"2026-10-07","author":{"@id":"https://vpnrank.io/#editorial-team"},"publisher":{"@id":"https://vpnrank.io/#organization"},"inLanguage":"en-US"}
```