VPN BENCHMARK REPORTS · EVIDENCE IN CONTEXT

VPN Benchmark Week — 2026-09-07

Explore the published provider and protocol measurements for 2026-09-07. Read speed retention alongside sample counts, destination coverage and latency before interpreting a difference.

YOUR NEXT STEP

The reporting window comes first

An ongoing period may gain more data. The source window shown below describes the measurements currently included, not a guarantee about your own connection.

THIS REPORT AT A GLANCE

The evidence behind this period

Published window2026-09-07 → 2026-09-13UTC where specified by the source

Measured setups6Provider × protocol rows

Providers in this report2Coverage differs by setup

NordVPN

ExpressVPN

Published source context: Data-driven report generated from QC-rankable benchmark measurements between 2026-09-07 and 2026-09-13. It describes observed results rather than claiming one provider is universally best.

PROVIDER × PROTOCOL RESULTS

Inspect the measured setups

Filter by provider or search for a protocol. Each row describes a measured setup; read its sample and destination counts before comparing the median.

ProviderFind a setup

6 published setups

VPN Benchmark Week — 2026-09-07 — published measurements
Provider	Protocol	Samples	Destinations	P10	Median	P90	Download	Latency
NordVPN	NordLynx	2,155	51	37.5%	65.8%	100.0%	48.2 Mbps	132.4 ms
ExpressVPN	Lightway UDP	4,615	50	22.3%	64.4%	100.0%	46.5 Mbps	112.9 ms
NordVPN	OpenVPN TCP	794	56	27.6%	56.6%	93.1%	46.7 Mbps	153.2 ms
ExpressVPN	OpenVPN UDP	866	51	25.8%	54.4%	100.0%	46.0 Mbps	119.1 ms
NordVPN	OpenVPN UDP	813	54	29.5%	53.2%	94.2%	42.3 Mbps	126.1 ms
ExpressVPN	OpenVPN TCP	837	54	6.7%	16.6%	64.6%	14.7 Mbps	119.8 ms

Figures are reproduced from the report source without recomputing or reranking them. A destination count does not identify the countries included.

MAKE THE FIGURES USEFUL

What should you look at first?

Start with the window

Read the UTC reporting dates. A weekly report can capture a short-lived change; a monthly report can smooth it out.

Read the spread

The median is the midpoint. P10 and P90 show how widely the measured results vary; a single headline misses that variation.

Check the coverage

Sample totals and destination counts differ between setups. Pooled report rows do not automatically form a matched-route comparison.

KEEP THE COMPARISON FAIR

A larger median is a question worth exploring

Different routes, source locations and sample mixes can produce different aggregates. Use the table to identify a setup to investigate, then move to a matched-source comparison.