LeoTun

Internet Freedom Alliance

2026-10 sing-box upgraded: old configs no longer work. Sign in and re-scan the code on the sing-box page, once per device.
2026-09 New router firmware is out — please update. OpenVPN is back: download a profile, import it, and connect in one tap.
2026-09 New Personal plan: 5 USD/month for 2 devices, routers included — everyone should have a VPN router.

Related guides

VPN Slow? How to Run a Speed Test and Find the Cause

Updated

The short answer: with a VPN on, data takes a longer path and goes through one more layer of encryption, so it is always a little slower than the bare line. When it is much slower, one of five segments is holding it back: home Wi‑Fi, the local broadband, the cross-border link, how busy the chosen region is, or the device and protocol. A fair speed test starts without the VPN as a baseline, then repeats against the same test point with the VPN on, changing one thing at a time. This guide covers why a VPN slows down, how to test so the result means something, how to read the numbers, how to find the slow segment step by step, and what to change.

Why a VPN is slower than the bare line

  • Distance: data goes to the VPN server first and then to the site, so the path is longer and latency rises. The further the server is from you and from the site, the bigger the detour.
  • The cross-border link: from inside China the narrowest segment is usually the carrier’s international exit. The same account can differ several-fold between regions because of the carrier’s line to that region, not the servers.
  • Peak hours: between 8 and 11 pm everyone is online at once, and home broadband, international exits and servers are all more crowded.
  • Home Wi‑Fi: walls, distance from the router and neighbouring networks can make the Wi‑Fi segment far slower than the broadband itself. It has nothing to do with the VPN but often gets blamed on it.
  • Device performance: encryption and decryption are done by the device’s processor. On an old phone, a TV box or an entry-level router, the device itself becomes the ceiling.
  • Protocol: some networks restrict UDP and throttle the methods that use it; on lossy networks a loss-tolerant protocol has the edge. The same line can give very different results with a different connection method.

How to test so the result means something

  1. Turn the VPN off and measure local speed first, with speedtest.net, fast.com or your carrier’s own test page. Note download, upload and latency. This is the ceiling; no VPN can exceed it.
  2. Connect the VPN and test again, choosing a test server at the far end of the route: from inside China to an overseas region, pick an overseas test server, because a domestic one cannot measure the cross-border link; from abroad to the China region, pick a test server in China.
  3. Confirm you are measuring the route: the test page shows the IP and provider it sees, which should be the exit end. With split routing on, or with the web proxy, the test site may be sent direct and you end up measuring home broadband. Add the test site to the rules or switch to global for the test.
  4. Change one thing at a time. To compare two regions, keep the same test site, test server and device. To compare two times of day, keep the same region.
  5. Test once wired and once on Wi‑Fi. The wired result on a computer reflects the broadband and the route; if Wi‑Fi in the same spot is much lower, Wi‑Fi is the slow part.
  6. Test once in the daytime and once at 9 pm Beijing time. A large gap points to peak congestion, not a setting.
  7. Run each case two or three times and take the middle value. Single results vary.

How to read the numbers

  • Download (Mbps): what video, web pages and file downloads depend on. Mind the unit: divide Mbps by 8 to get the MB/s a download manager shows, so 40 Mbps is about 5 MB/s.
  • Upload: matters for video calls, sending files and live streaming, much less for ordinary browsing.
  • Latency (ms): the time between a click and a response. One or two hundred milliseconds is common once traffic crosses a border. Games and calls are sensitive to it; video is barely affected.
  • Jitter and packet loss: latency that swings up and down, and packets that never arrive. Choppy calls and frequent buffering usually come from these, not from a low download figure.

How much speed video really needs

Going by the figures the streaming platforms publish: about 3 Mbps for 720p, about 5 Mbps for 1080p, and 15 to 25 Mbps for 4K. A steady result in the low teens of Mbps is already enough for 1080p; there is no need to chase a large number.

If the number is high enough and video still stutters, look at stability: packet loss, jitter and time of day. Dropping the resolution one step to confirm smooth playback, then raising it again, tells you more than staring at a speed test.

Finding the slow segment, step by step

  1. VPN off, computer on a cable. If the result is far below what your broadband plan states, the problem is the broadband or the modem. Restart the modem and router, and contact the carrier if it persists.
  2. VPN off, same spot, on Wi‑Fi. If it is much lower than wired, the problem is Wi‑Fi. Move closer to the router, join the 5 GHz band and test again.
  3. VPN on, two or three different regions. If the regions differ a lot, the problem is your carrier’s exit to that region. Choose by the carrier list in Which region is fastest from inside China.
  4. Same region, daytime and evening peak. If only the evening is slow, it is congestion; pick a region showing a green light on the region list.
  5. Same region, a different connection method. If that is faster, the first protocol is restricted on this network; keep using the faster one.
  6. A different device. If the phone is fast and the TV box or router is slow, that device’s performance is the ceiling.
  7. Only one site or app is slow while the speed test is fine: usually that site’s own servers, or split-routing rules sending it down the other path. Switch to global once to compare.

What to change for each cause

  • Slow broadband or Wi‑Fi: fix the local side first. Put TVs and desktops on a cable where you can, place the router in the open, and restart a modem that has been running for months.
  • Wrong region: choose by carrier — northern Unicom try Japan and Korea first, southern Telecom try Southeast Asia or Australia, Mobile and other broadband try the China entry. Details are in the region guide.
  • Evening peak: pick a green-light region or switch connection method. Either helps more than reconnecting.
  • A network that restricts UDP (campus, office): switch to Cisco, which runs over TLS.
  • A lossy network (older community broadband, mobile data): switch to Hiddify, whose protocol is built to tolerate loss. Which connection method fits which scenario matches methods to networks.
  • An underpowered device: install a client directly on a faster device. For a whole household, see Choosing a VPN router.
  • Still slow after all of this: bring the method, region, a speed-test screenshot and the time to support. The checking order is in Cannot connect, slow, or dropping.

Bilibili and other Chinese video buffering abroad

When you watch Chinese video from abroad, the stutter usually comes from the cross-border segment: the video travels from servers in China to you, the path is long, and China’s evening peak makes it more crowded even when it is daytime where you are. Check these points one by one.

  • Use a whole-device method (Cisco, Hiddify, private network, OpenVPN or the router) with the China region. The web proxy handles pages in the browser, and video playback does not always go through it.
  • From abroad, run the client in global mode. Split routing is for users inside China.
  • For a speed test, pick a test server in China so you measure the leg back to China.
  • Try again outside 8–11 pm Beijing time. A big difference means peak congestion.
  • Drop the resolution one step to confirm smooth playback, then raise it.
  • A region or copyright message is a different problem: see Tencent Video or iQIYI blocked abroad.

FAQ

Which tool should I use for a VPN speed test?

Three common ones cover it. speedtest.net lets you choose the test server, which suits comparing regions. fast.com measures download speed to video servers, closer to what streaming feels like. speed.cloudflare.com also reports jitter and packet loss. Stick to one when comparing.

The speed test is fast but video still buffers. Why?

A speed test measures a short burst to the test server; video needs a steady flow from the video servers. Common causes are packet loss and jitter, the evening peak, or split-routing rules sending the video site down the other path. Try a green-light region, and try global mode once.

Can a VPN ever be faster than no VPN?

Occasionally: when your carrier’s direct route to a particular site is poor, the detour through the VPN can be smoother. Most of the time a VPN is slightly slower than the bare line, and that is normal.

Why is every speed test result different?

Other users on the same line, other devices at home, Wi‑Fi interference and the load on the test server all keep changing. Run two or three tests, take the middle value, and keep the test site, server and device fixed so results can be compared.

Should I switch to another server when it is slow?

Switch region; there is no need to pick machines. Each region has several machines and the system picks the least busy one at connection time. For more speed, a green-light region is the least effort.

The router tests slower than the client on my phone. Is that normal?

The router’s processor may be the ceiling: it does the encryption for the whole household’s traffic. Test the same region with the client on a phone; a large gap means device performance is the cause.

Related pages