V2Ray Slow or Unstable? 8 Practical Tips to Speed It Up

Most "slow" or "keeps dropping" issues don't need any advanced tinkering. Check these four things in order first, and you'll usually notice a real improvement within ten minutes.

1. Run a batch speed test — don't just use the first node: The client's batch speed test shows real latency for every node. Lower-latency nodes are usually faster and more stable, so test first and rotate through the top few.
2. Sync your system clock: The VMess protocol is very sensitive to clock drift — anything over 90 seconds can cause frequent disconnects or a total failure to connect. Enabling automatic network time sync fixes this at the root.
3. Use PAC mode instead of Global Proxy: PAC mode only routes traffic bound for services outside mainland China through the proxy, connecting mainland China sites directly — cutting out unnecessary detours. In practice, it's usually faster and lighter on bandwidth than a global proxy.
4. Remove battery optimization restrictions on the client: On mobile, if the system's battery-saving policy restricts the client from running in the background, you'll see it disconnect on its own after a while. Adding the client to the battery optimization whitelist fixes this.

If speed is still lacking after the four steps above, the issue is often in your protocol choice or a poorly maintained subscription. Keep checking these:

5. Prefer VLESS + Reality or Trojan nodes: These combinations offer stronger censorship resistance and are less likely to be throttled in restrictive network environments. If your provider offers multiple protocols, try these first.
6. Try switching transport protocols: The same protocol can run over TCP, WebSocket, gRPC, or mKCP transports. If your current network is throttling one transport heavily, switching to another can often improve speed immediately.
7. Update your subscription regularly: Providers constantly maintain and rotate their nodes — going too long without updating your subscription can leave you stuck with dead or degraded nodes. We recommend clicking "Update Subscription" manually every few days.
8. Keep your client up to date: v2rayN, v2rayNG, and v2flyNG are all continuously updated — new versions usually fix known connection issues and improve core performance. Plenty of "mystery problems" simply disappear once you update to the latest version.

There's one detail that's easy to overlook — routing rules. Most clients support custom split-routing rules, and if mainland China traffic is mistakenly routed through the proxy too, it not only slows things down but also wastes your node's bandwidth. Check your routing / split-tunneling settings to make sure mainland China domains and common apps connect directly, and only traffic that genuinely needs the proxy goes through the node. The same logic applies to per-app proxying (common on Android) — only enable it for the apps that truly need it, rather than letting every app compete for the same node's connections and bandwidth.

If the problem persists after going through everything above, it's most likely a service quality issue on the node itself — oversold bandwidth, a poor-quality data center route, or too many concurrent users can all cause a "nothing I do helps" situation.

At that point, the most effective move is to reach out to your provider directly, or switch to a different node that performs better in your speed tests. The client configuration usually isn't the bottleneck anymore, so further tweaking will just waste your time.

Rather than coming back to this article every time something feels off, build these two small habits into your routine: first, run a batch speed test before each session — node performance fluctuates across different times of day and network conditions, and testing only takes a few seconds but saves you from sticking with a node that's no longer the fastest; second, update your subscription and check for client updates on a regular schedule, tying both to the simple act of "opening the client" so you eventually do it without even thinking about it.

Once these two habits stick, you'll find that most "sudden slowdowns" get caught early, long before they actually affect your usage — and your overall experience will be far more stable.