Speed is always a top concern for those shopping for a VPN, and yet it’s one of the most difficult factors to accurately quantify. We always run speed tests as empirically as possible when we review a VPN provider, but the fact of the matter is that the fastest VPN for where you live is not necessarily the fastest VPN for where I live. The quickest for streaming video might not be the speediest for online gaming. Even the fastest VPN at noon probably isn’t at midnight.
We always advise readers to take speed tests with a big grain of salt. Too many variables are at play. And we’ve also stated as clearly as possible, there is no such thing as the “fastest VPN,” no matter how many companies make such a claim.
All that being said, some VPNs are still all around faster than others. Below we’ve listed our top five fastest VPNs tested in the last year. While the speed tests we run factor largely into this list, other non-quantifiable parameters based on our personal experience are also taken into consideration. These include how well they stream HD video and game online.
ExpressVPN sets the bar when it comes to download speed. It’s always near the top of the rankings, albeit never at the peak. Consistency is a defining factor of Express; volatility is rarely an issue that affects the outcome of test results. Connections drop a little more often than we’d like them to, but the company has done a remarkable job considering the size of the network it manages. ExpressVPN offer a risk free 30 day money back guarantee and are currently running an offer of 3 free extra months with their 12 month plan here.
Read our full review of ExpressVPN here.
NordVPN also nudged out ExpressVPN in terms of speed, with a few caveats. Most VPN apps select a location, and then the app automatically selects the best server in that location. NordVPN is not so good at this. The auto-select on a couple occasions put us on servers that were complete duds, which resulted in a test result so bad it qualified as a statistical outlier and had to be thrown out. Thankfully, the app allows you to manually select a specific server and view the load capacity on all servers, where we had much better luck. Servers are optimized for specific streaming channels, torrenting, or security measures. Update, May 19, 2017: NordVPN is running a deal with a huge 72% off its 24 month plan. It is a little hidden away on the site and we are not sure how long it will run for, full details can be found on NordVPN’s site here.
Read our full NordVPN review here.
A relative newcomer, LiquidVPN impressed us with a sophisticated overall package and excellent download speeds. It narrowly beat ExpressVPN in our tests, but was also a bit more volatile. The company invests in Tier 3 data centers, which is the second-most reliable option. Tier 4 is usually only reserved for mission-critical processes and data. Some servers are optimized for peer-to-peer applications (read: torrenting), and others are DDoS protected to prevent dropped connections. You can try it with a 7 day money back guarantee here (add the discount code “COMP20” at checkout for a 20% discount).
Read our full review of LiquidVPN here.
Buffered might not be quite as quick as some of the other VPNs on this list, but it’s stable as a rock. You get a steady download stream that isn’t as prone to volatile spikes and dips. With five simultaneous connections, we highly recommend it for a family or small office. The lack of dedicated mobile apps might be a deal breaker for some, though.
Read our full review of Buffered VPN here.
VyprVPN is one of the few providers that owns all of its own server infrastructure rather than just renting out space on someone else’s hardware. That means fast, consistent speeds that aren’t as affected by unrelated network traffic, with servers all over the world. We wouldn’t recommend torrenting on VyprVPN as it’s against the company’s policy, but it’s more than enough for streaming and gaming. You can get a 3 day free trial followed by your first month’s service half price here.
Read our full review of VyprVPN here.
How we test VPNs for speed
To test download speeds, we downloaded the same 82.7MB file (compressed from 103MB) three times from three different servers at three different times per day for a total of nine tests. For comparison’s sake, we also run the downloads without a VPN as a control group. When all is said and done, we plot the data in like this:
Tests are run on a 10Mbps connections. Our testers are in different locations around the world, but we never plot the results of speed tests from different locations together. The internet in Buenos Aires will almost always be slower than in Berlin, for instance, so that wouldn’t be a fair comparison. We do add multiple tests to the same plot even if we ran them days or weeks apart, so long as they are from the same location.
Update: We’ve received some feedback that 10Mbps is too slow to get good test results. We would argue that 10Mbps is quite average for home internet across the world, and we run tests for average people. While our tests might not be a good indication of speed if you’re paying for a 30Mbps or 50Mbps connection, the results are varied enough to get a statistically significant indication of overall performance. Furthermore, all the VPNs on the list officially offer unlimited bandwidth, so if you happen to be on an uncongested nearby server, chances are you’ll still be able to max out your available bandwidth. Finally, there is no point in having a fast VPN if it is unstable, doesn’t protect your privacy, doesn’t unblock the content you want, or doesn’t have a good range of servers to connect to. We excluded providers like TotalVPN which were very fast but had awful customer service, for example.
We do not test latency, as this is more an indicator of our location than the VPN’s speed. We also do not test upload speed.
Why we don’t use Speedtest.net
Many VPN reviewers out there still rely on Speedtest.net to measure download and upload speeds, but Speedtest.net is unfortunately not suitable for testing VPN speed. It does not account for LZO compression, which is built into the OpenVPN protocol–the only protocol we test.
This leads to some inexplicable results when using Speedtest.net. For example, we’ve seen speed tests showing faster download rates than what our ISP even allows.
Thus, the best option is to use a much less sophisticated yet much more reliable test, as outlined above.
Stability trumps speed
In our humble opinion, the primary performance factor when choosing a VPN shouldn’t be speed. It should be stability. More often than not, it’s volatility that brings down test scores rather than a slower overall download rate.
Every VPN will occasionally have a bad day or just a few bad hours where service is slow on a particular server or set of servers. Some VPNs have more high traffic periods or downtime than others. These are the ones to be avoided. Unfortunately, the test period for our reviews rarely lasts more than two weeks, so it’s difficult to predict what VPNs will encounter more issues in the long term at the time of writing.
That being said, the boxplots we provide in our reviews can give some indication of how volatile a VPN’s performance is. A larger blue box means more volatility, even if the mean and median download times are low, a lot of volatility will probably cause more frustration than a slightly slower connection.
Unless you’re on a fast (100+Mbps) connection, the chances of maxing out the bandwidth available is pretty slim. Almost all VPNs these days advertise unlimited bandwidth.
Speed shouldn’t be your only consideration when choosing a VPN. Depending on what you plan to use it for, you may want a VPN that’s best for torrenting or unblocking Netflix or gaming. Those aren’t necessarily the fastest, and they don’t need to be. They just have to be fast enough.
How to choose the fastest VPN server
You can choose the fastest VPN server from any provider based on the following criteria:
- Proximity: Choosing a server close to your own location will result in a speedier connection. If you’re in Asia and want to access American content, then choose a server on the west coast USA. If you’re in Europe, then use an East Coast server. If you know the location of the server hosting the content you’re trying to access, you can also choose a VPN server near there.
- Latency: This is closely tied to proximity, but is also affected by the amount of traffic on the networks between you and the VPN server. Latency measures the time it takes to send and receive a request from a server, also called ping time. Many VPN apps will allow you to see which server offer the least latency, usually measured in milliseconds. If not, you can connect to the server and use a terminal or command prompt to ping a website and view the time.
- Server load: A few VPNs, such as IPVanish and NordVPN, allow you to see the current load of VPN servers as a percentage of their total capacity. A larger server load means more traffic and slower connections.
- VPN protocol: We always recommend users avail of OpenVPN when available, as it is the most secure and open-source protocol available. You may, however, opt for a speedier protocol. IKEv2 is secure and works well especially with mobile data connections. You might or might not notice a difference with L2TP/IPSec or SSTP, depending on your device hardware. PPTP is generally regarded as the fastest, but has known security flaws that make it unsuitable for anyone who values their privacy.
Choosing the fastest server for gaming
If you’re an online gamer who uses a VPN to access another region’s servers (or because you got IP banned), the most important factor in choosing a VPN is latency. The ping time between the game servers and your computer or console is mostly what determines how much lag you’ll experience. If you want to stay competitive, figure out where the game’s regional servers are hosted and choose the nearest VPN server.
Download rate is also important, but gaming requires less bandwidth than you might think. The first priority should always be reducing latency.
Security versus speed
Adding security to a VPN connection inevitably results in a loss of speed. Using a stronger encryption algorithm, for example, means it takes longer to encrypt data travelling through the VPN and longer to decrypt it once it arrives at its destination. Similarly, more secure VPN protocols tend to be slower than less secure ones. PPTP, despite being the oldest protocol, is still significantly faster than OpenVPN or L2TP/IPSec. However, it also has known security vulnerabilities.
You don’t necessarily need the strongest available security all the time when using a VPN, but there’s a line that shouldn’t be crossed in order to increase speed. We recommend against using PPTP and advise readers to opt for a protocol that uses SSL (OpenVPN) or IPSec encryption (L2TP, IKEv2).
Most VPNs don’t give you the option, anyway, but don’t disable encryption altogether. Additionally, 128-bit AES is the minimum strength encryption necessary for a VPN to do its job and keep your data safe. It’s effectively un-crackable and is slightly faster than 256-bit AES, which is also common. A handful of VPNs use Blowfish encryption, which tends to be slower than its AES counterpart. We recommend at least 448-bit Blowfish encryption if you go that route.
How much of a speed boost you’ll get out of tweaking encryption and VPN protocols depends on your hardware. More powerful devices will be able to encrypt and decrypt data more quickly.
UDP versus TCP
When using the OpenVPN protocol, many VPN apps will give you the option of using either the TCP or UDP protocol. TCP is the protocol that you normally use to browse the web and download files. UDP is more common for streaming applications such as video, music, and gaming. The difference is in how computers and servers send network packets, the unit of data used to exchange information over the internet.
A common misconception is that UDP is always faster than TCP. While this is often the case, it’s more of a guideline than a rule.
TCP uses error protection and guaranteed delivery to ensure that every data packet is identical to the original and sent in the correct order. If there is an error, the flow of internet traffic is stopped until the previous packets are sent successfully. This is important if you’re downloading a file or loading a web page.
UDP doesn’t have these protections. Collisions, errors, and missing packets are all common, and sometimes packets arrive out of order. The emphasis is on speed, not being perfect. It is preferable to miss a packet than delay all of the other packets following it.
So, if you use your VPN primarily for online gaming, streaming video, or listening to music, then switch to UDP. Otherwise, stick to TCP.
VPNs secure your traffic and route it through an intermediary server so it can’t be traced. But if privacy is not of chief concern to you, then there are other alternative proxy methods that offer faster speed. A SOCKS proxy, for example, does pretty much the same thing as a VPN without the encryption. Without having to encrypt and decrypt traffic, SOCKS proxy users can get faster speeds and still mask their IP address.
Then there’s smart DNS proxies. These proxies only re-route your DNS requests to make it appear as though you are in a different location, rather than all of a device’s traffic. This means you still get the full benefit of a direct internet connection, but not the privacy or security of a VPN.
Avoid “speed boosters”
Some Android apps, iOS apps, and desktop browser extensions claim they can speed up your VPN connection. These are almost undoubtedly scams.
There is simply no way for a third-party app to increase the speed of your VPN connection beyond what we’ve outlined above, especially not for free. More likely, these apps are used to mine your data or serve ads.
The closest you might get to a real speed booster is to subscribe to an optimized gaming network, which ensures that your internet traffic is taking the shortest and least-congested route available to give you a better ping time. But these are paid subscription services, not free apps, and even then their effectiveness is questionable.