How Fast Is Your VPN?
Measure real download speed, upload speed, ping and jitter through your VPN connection.
Test History (This Session)
| Time | Download | Upload | Ping | Jitter | Rating |
|---|---|---|---|---|---|
| No tests run yet | |||||
Why a VPN Slows Your Connection Down
Every VPN costs you some speed. That is not a defect - it is the unavoidable price of routing your traffic through an extra machine and encrypting it on the way. Three separate factors contribute, and knowing which one dominates tells you what to change.
- Distance: your packets travel to the VPN server before reaching their destination. A server on another continent can add 150-300 ms of latency no matter how fast your connection is, because the limit is the speed of light in fibre.
- Encryption overhead: every packet is encrypted and decrypted at both ends. On modern hardware with AES acceleration this costs little, but on older routers and low-power devices it can become the bottleneck.
- Server congestion: the most common cause of poor VPN speed by a wide margin. A popular server shared by thousands of users will be slow regardless of your own connection or the provider’s marketing claims.
A useful rule of thumb: a well-configured VPN on a nearby server should cost you 5-15% of your download speed. Losses above 30% almost always point to a congested server or a poorly chosen protocol rather than to any fundamental limitation.
How to Read Your Results
Four numbers are measured. Download speed gets all the attention, but for anything interactive, ping and jitter matter more.
| Metric | What it measures | Good value |
|---|---|---|
| Download | How fast data reaches you. Determines streaming quality and download times. | 25+ Mbps for 4K video; 5 Mbps for HD |
| Upload | How fast you send data. Matters for video calls, backups and file sharing. | 10+ Mbps for smooth video calls |
| Ping | Round-trip latency. The delay before anything begins to happen. | Under 50 ms is good; under 100 ms is usable |
| Jitter | Variation in ping between samples. High jitter means an unstable connection. | Under 30 ms; under 10 ms is excellent |
Jitter is the metric most people ignore and the one that ruins calls. A connection with 200 Mbps download and 80 ms of jitter will produce choppy audio and frozen video, while 20 Mbps with 5 ms of jitter will feel perfectly smooth. If your calls stutter on a fast connection, jitter is almost always the reason.
To measure the real cost of your VPN, run the test twice: once disconnected, once connected to the server you normally use. The difference between the two runs is your actual overhead. A single connected measurement tells you very little on its own, because it cannot distinguish a slow VPN from a slow line.
How to Make Your VPN Faster
In rough order of how much difference each change tends to make:
- Switch to WireGuard. It is dramatically faster than OpenVPN on most connections - frequently two to three times the throughput - because of a leaner codebase and modern ciphers. If your provider offers it under another name (NordLynx, Lightway), that counts.
- Choose a closer server. Latency is bounded by physics. A server in your own country will nearly always beat a distant one, whatever its advertised capacity.
- Avoid congested servers. Many clients display a load percentage; anything above 70% is worth avoiding. Trying three nearby servers and keeping the fastest takes a minute and often doubles throughput.
- Use a wired connection. Wi-Fi adds latency and jitter that no VPN setting can compensate for. This matters more for calls and gaming than for downloads.
- Turn off double VPN and Tor-over-VPN. These features route traffic through two or more hops and cost most of your speed. They are worth it only for a threat model that genuinely requires them.
- Check for throttling. If your VPN is faster than your unprotected connection for a given activity, your ISP is likely throttling that traffic - which is one of the few cases where a VPN legitimately makes things quicker.
How This Speed Test Works
The test runs entirely in your browser and measures the connection as it exists at this moment, through whatever tunnel is currently active.
- Ping: a series of small requests is timed, and the median round-trip is reported to avoid a single outlier skewing the result.
- Jitter: calculated as the mean deviation between consecutive ping samples - the standard measure of connection stability.
- Download: payloads of increasing size are fetched, and throughput is measured once the transfer reaches a steady state, so TCP slow-start does not depress the figure.
- Upload: generated data is posted and timed using the same steady-state approach.
Browser-based tests are bounded by what a browser can do. On very fast lines - above roughly 500 Mbps - the JavaScript engine and single-connection limits become the constraint rather than your network, so treat high-end results as a floor rather than a ceiling. Full details are on our methodology page.
🧰 Related VPN Tools
Every test runs entirely in your browser - nothing is stored unless you choose to share the result.
VPN Speed Test - FAQ
How much speed a VPN should cost you, what the numbers mean, and how to get more of them back.
How much speed should I lose with a VPN?
On a nearby server with a modern protocol, expect to lose 5-15% of your download speed. Losses of 20-30% are normal on distant servers. Anything beyond that usually indicates a congested server, an outdated protocol such as OpenVPN over TCP, or hardware that cannot keep up with the encryption.
Why is my VPN so slow?
Server congestion is by far the most common cause, followed by distance to the server and protocol choice. Try three different nearby servers and switch to WireGuard before concluding your provider is slow - those two changes resolve the majority of complaints. If every server is slow, test without the VPN to establish what your line is actually capable of.
Which VPN protocol is fastest?
WireGuard, by a clear margin. Its implementation is a fraction of the size of OpenVPN’s and it uses ChaCha20-Poly1305, which performs well even on devices without AES hardware acceleration. Providers ship it under various names - NordLynx, Lightway and similar are WireGuard-derived. IKEv2 is a reasonable second choice; OpenVPN over TCP is the slowest option in common use.
What is jitter and why does it matter?
Jitter is the variation in latency between consecutive packets. Steady 60 ms latency feels fine; latency swinging between 20 and 200 ms does not, because real-time applications must buffer for the worst case. It is the metric that determines whether video calls and gaming feel smooth, and a fast connection with high jitter will perform worse than a slower, stable one.
Can a VPN make my internet faster?
Occasionally, yes. If your ISP throttles specific traffic - streaming or peer-to-peer are the usual targets - a VPN hides the traffic type and removes the throttling, which can produce a genuine improvement. A VPN can also sometimes route around a congested peering link. Both are exceptions; a VPN adds overhead by default.
Is this test accurate on gigabit connections?
It is accurate up to roughly 500 Mbps. Beyond that, browser limitations rather than your network become the constraint, so results understate what the line can do. For gigabit measurement, a native speed-test client will give a truer figure. For comparing VPN on versus VPN off - which is what this page is for - the relative difference remains meaningful.
Should I test with the VPN on or off?
Both, one after the other. A connected measurement in isolation cannot tell you whether a low number comes from the VPN or from your line. Test disconnected first to establish a baseline, then connected to the server you normally use. The gap between the two is your VPN’s real cost, and it is the only number worth comparing between providers.
Why do results vary between runs?
Network conditions change constantly - server load, routing, local Wi-Fi interference and other devices on your network all shift from minute to minute. Variation of 10-20% between consecutive runs is expected. Run the test three times and use the median; the session history table above is there so you can see the spread rather than trusting one sample.