VPN Connected but No Internet (June 2026): Three Different Causes and How to Tell Them Apart in 2 Minutes
Short answer: if your VPN shows "Connected" but pages won't load, you almost certainly have one of three different problems — and they are fixed in different ways. Since late May and early June 2026, three mechanisms have stacked up in Russia: a new "freezing" filtering scheme (the TSPU stopped *cutting* connections and started *stalling* them), mobile-internet shutdowns based on "whitelists" (where a VPN cannot help at all), and soft speed throttling by carriers. The symptom is identical for all three — "connected but useless" — so people blame their config or app, when the cause often isn't there at all.
Below is how to figure out which one you have in two minutes, and what actually works in June 2026.
First — a 2-minute diagnosis
Before changing any configs, run three checks. They immediately rule out half the options.
1. Open a Russian site without a VPN (e.g. gosuslugi.ru or ya.ru), then a foreign one (google.com). If Russian sites open but foreign ones don't even *without* a VPN — that's a whitelist (cause #2), and a VPN is powerless here.
2. Compare Wi-Fi and mobile data. Turn off Wi-Fi, try the VPN on mobile, then the other way round. If it works on one but not the other, the problem is the carrier/network (cause #1 or #3), not the app.
3. See whether the connection drops or "hangs". A VPN disconnecting with an error is one thing. A VPN that stays "connected" while the page loads forever and sometimes "unfreezes" after 1–2 minutes — that's the new June scheme (cause #1).
Remember your result — we'll break down each cause next.
Cause #1: the new TSPU scheme — the connection "freezes" instead of dropping
This is the big news of June 2026 and the reason for the wave of complaints. Previously, when the filtering system (TSPU) detected a tunnel, it simply cut the connection — you saw an error, swapped configs, moved on. Now the approach has changed: instead of dropping it, traffic to a suspicious address is slowed almost to zero for about 2 minutes. The connection is technically alive, the app says "Connected", but no data flows. And if you start poking at settings at that moment (changing the fingerprint, reconnecting), the system flags "adaptive behavior" and extends the freeze for much longer.
A technical breakdown on Habr (June 6, 2026) describes that the block now triggers on a match of several signals at once:
- the server address sits in a "suspicious" subnet — mass data centers like Selectel, Yandex Cloud, Hetzner, DigitalOcean, OVH;
- a TLS fingerprint that looks like Chrome/Safari/iOS (the default for most ready-made configs);
- several parallel connections to the same address within a fraction of a second.
When all three line up, the traffic freezes. That's exactly why "free configs from public channels" now last only hours: they are built on cheap mass VPS from precisely these subnets and with a stock fingerprint.
This is confirmed at the state level too. According to an RBC breakdown (around June 10), Roskomnadzor has officially shifted to blocking "by indirect signals": IP ranges, TLS characteristics, connection frequency, fingerprints. The side effect — ordinary sites hosted on the same Russian providers go down along with the tunnels. And Rostelecom announced in early June a 1.3-billion-ruble purchase of servers for the TSPU — the official reason in the documents is a shortage of filtering capacity, because VPN downloads grew 14x year over year. In short: the pressure will only increase.
What actually helps against the "freeze"
Over June the community has found working tactics (more on protocols in Why VLESS Stopped Working in Russia):
- Change the TLS fingerprint from
chrometofirefoxoredge— removes one of the three triggers. - Move off mass data centers. A server outside Selectel/Yandex Cloud/Hetzner/DO/OVH is blocked noticeably less often.
- Use a modern transport instead of "bare" VLESS+Reality: XHTTP (including via the Cloudflare CDN) or Hysteria2 change the behavioral signature the freeze relies on.
- Don't poke settings under a block. If you've hit a freeze, reconnecting and changing parameters only prolongs it. Give it a pause.
- Keep the client on a current version (current Xray-core), but without overdoing it: some fresh builds are buggy — the community, for example, temporarily rolled back from problematic v2rayNG versions to stable ones.
The core principle of June 2026: adaptation wins, not a single baked-in config. Any static protocol eventually falls under a new signature.
Cause #2: "whitelists" — a VPN can't help, and it's not your fault
The second cause looks the same ("connected but nothing loads") but has a completely different nature — and here a VPN is powerless by definition.
In some regions and in Moscow on restriction days, a "whitelist" mode is switched on: only traffic to approved Russian services (government portals, banks, VK, Rutube, domestic maps) is allowed, while everything else — non-Russian — is cut at the network level, before your VPN tunnel even reaches its server. On June 9–10, 2026 in Moscow there were around 2,500 complaints a day: sites, maps, and video wouldn't open, Telegram and YouTube were down — while VK and Rutube worked. That's the whitelist signature: "ours" works, "theirs" doesn't.
The official framing for such shutdowns is counter-drone measures (SIM cards are used in drone navigation), so it's mainly mobile internet that gets cut, often on a schedule. In some regions the restrictions are kept almost permanently.
How to tell it's a whitelist
- Foreign sites won't open even without a VPN, while Russian ones do.
- On home Wi-Fi (a wired provider) everything is fine, but on mobile it isn't.
- Everything foreign stops at once, not just one service.
What to do: during such shutdowns switch to Wi-Fi (wired providers are hit by whitelists less often), or simply wait it out — these are usually temporary windows. Changing configs is pointless: the problem isn't the tunnel.
Cause #3: throttling — "works, but unusable"
The third option: the connection is alive, sites even open, but the speed drops to a level where nothing works. Some carriers in 2026 apply a soft slowdown to recognized VPN traffic — for instance, to ~0.5 Mbps — without dropping the connection. Video won't load, pages take half a minute, and the VPN indicator honestly stays on "Connected".
How to recognize throttling
- The VPN is connected, simple pages open, but slowly; video and heavy content won't load.
- The problem is only on a specific carrier's mobile network; speed is normal on Wi-Fi.
- A speed test under the VPN shows abnormally low values specifically with the tunnel on.
What helps: a masking transport that doesn't reveal itself as a VPN (XHTTP/Reality with the right parameters, Hysteria2 with obfuscation), and a server outside the typical subnets. If the carrier doesn't "recognize" the traffic as a tunnel, there's nothing to throttle.
Summary table: telling the three causes apart
| Signal | #1 TSPU freeze | #2 Whitelists | #3 Throttling |
|---|---|---|---|
| Foreign sites without VPN | won't load / partial | won't load | load |
| Russian sites | work | work | work |
| On Wi-Fi | often works | usually works | works |
| On mobile | hangs ~2 min | doesn't work | very slow |
| Does a VPN help | yes, with a transport change | no | yes, with masking |
| What to do | change fingerprint/transport/server | wait / Wi-Fi | masking protocol |
Why "just download a new config" no longer works
The unifying point across all three causes: the era of "find a free key and sit on it for a year" is over. Free configs from public channels are built on mass VPS in "suspicious" subnets, with a stock fingerprint, and fall under the freeze within hours. Paid single keys last longer but are still static — when the TSPU shifts its detection method, such a key silently stops working and you're hunting again.
A separate June pain is the client crisis: V2RayTun has been officially declared unsupported and removed from the Russian App Store, Happ had a serious vulnerability with open access to configs found this spring, and Amnezia has been down since May 20 due to blocks and DDoS. People bounce between clients, but the problem isn't the client — it's the static servers behind it.
What works reliably: an adaptive service instead of a manual config
If you're willing to dig into SNI, fingerprints, transports and keep your finger on the pulse — great, the sections above give working tactics. If you'd rather have it "just work", you need not a single static config but a service that adapts server-side.
MegaV VPN runs the V2Ray/Xray stack on managed servers outside mass data centers, switches between modern transports (XHTTP, gRPC, current flows), and rotates parameters as TSPU methods change. The very "freeze" from cause #1 is handled exactly this way — on the service side, without your involvement. The app installs on phone, laptop, and tablet and keeps the connection alive on MTS, Beeline, MegaFon, and Tele2. There's a 3-day free trial to test connectivity on your own network before paying.
And if you already have a client you like (v2rayNG, Hiddify, sing-box, or an OpenWrt router with PassWall) and don't want to switch — MegaV offers a subscription link for your own client: paste it in, get our servers on current transports, and when filtering shifts again, just hit "update subscription".
One important caveat: under whitelists (cause #2) no VPN will help — that's a network-level traffic cut, not a protocol block. Only a wired-provider Wi-Fi or waiting out the window helps there.
Frequently asked questions
The VPN says "Connected" but there's no internet — is it a virus or a broken phone?
Almost never. In 2026 this is the result of network-side filtering, not a device problem. Run the diagnosis at the top of this article — it shows which of the three causes you have.
Why does the VPN work on Wi-Fi but not on mobile?
Because whitelist shutdowns and part of the DPI restrictions are applied first to mobile networks. Wired providers are hit by them less often.
I've already changed five configs — nothing helps. What's wrong?
If the symptom is a "freeze" (cause #1), frequently swapping configs only makes it worse: the system catches the adaptive behavior and extends the block. You need a different transport and a server outside mass data centers — or a service that does this for you.
Will Hysteria2 save me?
Often — yes, because the TSPU is tuned primarily for TCP. But there's a catch: on some carriers (e.g. MegaFon in Moscow) almost all UDP is blocked, and Hysteria2 won't come up there without obfuscation. The working scheme is Hysteria2 with obfuscation plus a fallback Reality/XHTTP where UDP is closed.
If this article helped you sort things out, take a look at what to do if V2RayTun isn't working, why VLESS stopped working in Russia, and how to set up Hysteria2.