Skip to content

Proton VPN: Swiss Privacy With a Free Tier

Proton VPN: Swiss Privacy With a Free Tier

Proton VPN on Android: why Work Profile apps can dodge split tunneling, and how kill switch vs Private DNS explains no-internet after disconnect.

App summary

App
Proton VPN
Developer
Proton AG
License
GPL-3.0-only
Version
unknown
Size
unknown
Price
Free tier available; paid plans vary by region and billing period, exact prices unknown

Proton VPN’s Android app arrives with an unusual combination of properties: a free tier that carries no ads, an Android client whose source code is published under GPL-3.0, and an operator, Proton AG, based in Switzerland. Those three facts do most of the work in the coverage that already exists. They also tend to paper over the parts of the Android experience that generate actual support questions — work profiles, kill-switch states, Private DNS settings, and what the free servers can realistically carry at 7 p.m. on cellular data.

This review sticks to what is documented and says so plainly when something is not. The version number, the APK size, the minimum Android release, and regional pricing are all unknown here. So are a handful of behaviors that matter more than they sound: how split tunneling treats work-profile apps, how the client coexists with Android’s Private DNS, and whether the GitHub source has a matching installable build. Those gaps get their own sections rather than a shrug.

Proton VPN Android Work Profile: Why Split Tunneling May Not See Every App

Split tunneling is one of Proton VPN’s listed key features, and on a personal phone the concept is simple enough: you choose which apps bypass the tunnel and which ones go through it. A phone enrolled in an employer’s device management is a different animal. Android separates that device into a personal profile and a work profile, each with its own app inventory and, in some cases, its own copy of the same app. Whether Proton VPN’s split-tunneling screen enumerates work-profile packages at all is not documented in the fact data. Mark that as unknown rather than assuming either answer.

The temptation is to treat the public source code as the resolution. The Android client is published on GitHub under GPL-3.0-only, so anyone who wants to can read it. But a public repository is not a support statement. Reading code tells you what a particular build did at a particular moment; it does not tell you that work-profile packages are enumerated correctly, that the behavior survives the next Android release, or that the company will treat a work-profile bug as in scope. Open source raises the ceiling on what can be verified. It does not, by itself, verify this.

Tier is a third variable. Some advanced features require a paid plan, and the fact data does not say whether split tunneling sits inside that group or whether its behavior is identical on free and paid. Anyone who needs a specific work app to stay outside the tunnel, or a specific one to stay inside it, should treat the whole scenario as unproven until they have checked the official documentation and tested with something non-critical.

There is a practical rule that survives all of this. If a work app must always be tunneled — a mail client, an internal dashboard, anything that fails closed — do not build your setup around a split-tunneling exclusion. Leave it inside the tunnel and accept the throughput cost. Exclusions are a convenience feature, and convenience features are the wrong place to put a dependency you cannot afford to lose.

Proton VPN Android No Internet After Disconnect: Kill Switch vs Private DNS Checks

The kill switch is a listed Proton VPN feature, and its entire job is to stop traffic when the tunnel is not up. That design has a consequence people rarely think about until it happens: if the kill switch remains engaged after you disconnect, the phone has no usable route to the internet. The symptom looks like a broken connection. It may simply be the feature doing exactly what it was built to do, holding the door shut until the state is cleared.

The second suspect is Android’s Private DNS setting, which lives at the operating system level and can be set to a hostname for DNS over TLS. The fact data does not describe how Proton VPN interacts with that setting. It does not say whether the client overrides it, respects it, or ignores it entirely. That interaction has to be labeled unknown, and the honest diagnostic sequence treats it as a separate check rather than a proven cause. A phone that resolves nothing after a disconnect could be sitting behind a stale tunnel rule, or it could be pointed at a DNS resolver it can no longer reach.

Order the checks instead of guessing. Confirm the VPN is actually disconnected rather than reconnecting. Look at the kill switch state before touching anything else. Then look at Private DNS and try the automatic setting. Then switch networks — Wi-Fi to cellular, or the reverse — because a captive portal or a flaky carrier route can produce the same symptom with the VPN entirely innocent.

That last point deserves emphasis. Performance depends on network conditions and server load, and the fact data says so directly. Not every post-disconnect outage belongs to the VPN. A carrier data session that has quietly died, a router that has stopped handing out leases, a DNS resolver having a bad afternoon — all of these look identical from the notification shade. The fact data also does not document a troubleshooting procedure, which means the official support channels at protonvpn.com are the right destination once the obvious checks are done.

Proton VPN Free on Android Mobile Data: Congestion by Time of Day, Not a Data Cap

There is a durable belief that free VPN tiers meter you, and it usually surfaces as a claim about a monthly data allowance. The fact data does not support that. What it lists as free-tier drawbacks are limited server locations and limited speed. It also states that the free tier is available without ads. A slowdown after a few gigabytes of use is not evidence of a cap, because no cap is documented. It is evidence that something else is going on.

What is documented is that performance depends on network conditions and server load. On mobile data, both of those variables move constantly. Cell towers hand off, sectors get busy, backhaul saturates, and the free server pool absorbs whatever demand shows up. Congestion by time of day is a plausible explanation for the pattern where the same server feels fine at dawn and useless during the evening commute. Exact server-load figures are unknown, so nobody can put a number on it, but the mechanism is consistent with the stated drawbacks.

The mistake to avoid is expecting paid-server behavior from a free connection. Some advanced features require a paid plan, and speed is explicitly listed as a free-tier limitation. If your reference point is a paid server at two in the morning, a free server at six in the evening will feel broken by comparison. That is not a bug report. It is the tier working as described.

You can get directional evidence without lab equipment. Try the same server at different hours and note whether the pattern holds. Try the same hour on Wi-Fi and on cellular. If both networks degrade at the same time of day, the server pool is the likely constraint. If only cellular degrades, your carrier is a candidate. Neither test proves anything conclusively, but both beat assuming a data cap that the documentation never mentions.

Proton VPN Android GitHub APK vs Play Store: Update Path and Unknown Version

The Android client source code is published on GitHub under GPL-3.0-only, and the official site is https://protonvpn.com. Those two facts get compressed, in a lot of discussion, into the idea that there is a “GitHub version” and a “Play Store version” and you pick one. Public source is not the same thing as a published, signed, installable build. The fact data does not state whether GitHub hosts release binaries at all, which means the comparison cannot be made from the available information.

Version numbers are equally unavailable. The current latest Android app version is unknown here, and so is the APK size. Without those, there is no basis for claiming that one channel is ahead of the other, that the builds are identical, or that a sideloaded package matches a specific commit. Any article that states a version number for this app is working from something other than the facts assembled here.

The update paths do differ in a way that matters regardless of build parity. A Play Store install updates automatically, in the background, on Google’s schedule. A sideloaded package does not. Someone who installs a build obtained outside the store has taken on the job of watching for new releases and reinstalling by hand, and the fact data does not document how that process is supposed to work for this app. That is an unknown, not a workflow.

The safe instruction is the boring one: use the official Proton VPN channels, and if you want to read the source, read it on GitHub as source. Do not treat third-party mirrors as equivalent, and do not assume that a repository being public means every distributed build corresponds to it. If build provenance matters to you — and for a privacy tool it reasonably might — ask Proton directly rather than inferring from a file listing.

Android Always-On VPN and Proton VPN: Where the OS and the App Overlap

Android ships an always-on VPN setting and, alongside it, an option to block connections that do not use the VPN. These are operating system controls, not app features, and they exist independently of whatever client you install. Proton VPN lists a kill switch among its key features. That gives you two layers that can both decide your phone should not have internet, and the fact data does not describe how they interact.

Whether the app’s kill switch and Android’s always-on toggle duplicate each other, conflict, or simply stack is unknown. The fact data is silent on the point. What can be said is that enabling always-on with blocking changes the failure mode of the whole device: if the VPN cannot connect, nothing else connects either. That is the intended behavior of the setting, and it is the same behavior that produces the “no internet after disconnect” complaint described earlier, just triggered deliberately.

Treat the toggle as a commitment rather than a convenience. Turn it on when you are willing to accept that a VPN outage means a phone that cannot reach anything until you intervene. Know where the setting lives before you need it, because the moment you need it is the moment you have no connectivity to look it up. And if you are troubleshooting a device that will not get online, check this setting early — it is a common reason a phone looks broken when it is behaving exactly as configured.

What the Public GPL-3.0 Android Client Actually Verifies

Start with the misconception this kills: the Android app is not closed source. The client code is published on GitHub under GPL-3.0-only. That is a real, checkable fact, and it distinguishes Proton VPN from plenty of competitors that ship opaque binaries and call it a security model.

What the publication verifies is narrower than it sounds. It verifies that the client-side code is inspectable by anyone who wants to look, including researchers with the time and skill to do it properly. It does not verify server behavior. Logging happens, or does not happen, on infrastructure that is not in that repository. It also does not verify that a given installed build corresponds to a given commit, and the fact data does not establish that correspondence either way.

The no-logs claim is a separate thing entirely. Proton AG operates in Switzerland and states a no-logs policy, and the fact data directs users to review the current privacy policy and independent audit reports for the details. That is the right instruction, and it is worth being clear about why: reading client code is not an audit. An audit examines the claims a company makes about its own systems, with evidence and scope and a report you can argue with. Code review examines a program. They are complementary, and neither substitutes for the other.

For an ordinary Android user, the practical value of the open client is that scrutiny is possible at all. You are unlikely to read the repository yourself. The point is that someone can, and that the license permits redistribution and inspection rather than forbidding it. That is a meaningful baseline. It is not a guarantee, and treating it as one is how people end up over-trusting a tool.

NetShield, Secure Core, and Tor Over VPN: Which Features the Fact Sheet Confirms

The confirmed feature list is short and specific: WireGuard and OpenVPN protocol support, NetShield ad and tracker blocking, Secure Core, Tor over VPN, a kill switch, and split tunneling. That is the inventory. Everything else you have read about this app’s capabilities is either derived from those items or invented.

What the fact data does not say is which of these sit behind the paywall. Some advanced features require a paid plan, and the sentence stops there. It does not name them. Anyone planning to subscribe specifically for Secure Core or Tor over VPN should confirm inclusion before paying, because the assumption that a headline feature is free is exactly the kind of assumption that produces refund requests.

NetShield is described as ad and tracker blocking, and it operates as a filter on the connection rather than as a browser extension. The fact data does not quantify how much it blocks or how often it breaks a site, and those numbers would vary by destination anyway. Secure Core and Tor over VPN both involve additional routing, and the fact data does not quantify the performance cost. Given that performance depends on network conditions and server load, the cost will not be a fixed number even if you measure it yourself.

One absence is worth naming because people ask about it constantly. Port forwarding does not appear anywhere in the fact data. Not for paid plans, not for free, not as a feature at all. If a workflow depends on an inbound port, this is not a documented solution, and no amount of forum confidence changes that.

Free Tier Expectations: Locations, Streaming, Torrenting, and Device Count

The free tier has limited server locations. That much is stated. The actual country list is not in the fact data, so any specific claim about which countries are available on the Android free plan is guesswork. Check protonvpn.com for the current list rather than trusting a screenshot from an old review.

Streaming is the other place expectations run ahead of documentation. The fact data does not list streaming support as a free-tier feature, and it does not list it as a paid feature either. Someone who needs Netflix US on the free tier should not install this expecting a result. The same applies to torrenting and P2P: neither appears in the feature list, and the free tier’s stated limitations are about server locations and speed, not about which traffic types are permitted. Absence from the fact data is not proof of prohibition, but it is certainly not a promise of support.

Device count is a third unknown that people fill in optimistically. The fact data does not state a simultaneous device limit for the free tier. It also does not state one for paid plans. Anyone planning around ten simultaneous devices is planning around a number nobody here can confirm.

If the free tier turns out not to fit, the alternatives named in the fact data are Mullvad and IVPN, both paid, both privacy-focused with open-source clients, and Windscribe, which offers a freemium model with configurable features. Each has its own tradeoffs, and none of them is a drop-in answer to the work-profile and Private DNS questions raised above.

Verdict

Proton VPN on Android is best understood as a privacy baseline rather than a performance product. If what you want is an encrypted tunnel from a Swiss operator, a client whose source is public under GPL-3.0, and a free tier that does not serve ads, the app delivers exactly that and asks nothing in return. It is a reasonable default for someone who wants their traffic wrapped on public Wi-Fi and does not need the connection to do anything clever.

It is the wrong tool for several specific people. Anyone who needs Netflix US on the free tier should look elsewhere, because streaming is not a documented free feature. Anyone who needs torrenting on the free plan is in the same position. Anyone who needs port forwarding for seeding will not find it mentioned at all. Anyone who wants ten simultaneous devices on free is working from a number the documentation does not supply. And anyone expecting paid-server speeds from free servers is going to spend their evenings filing complaints about congestion that the tier openly lists as a limitation.

Before committing to anything beyond casual use, resolve the unknowns for yourself. The version, the APK size, the minimum Android release, and regional pricing are all unstated here. So are the free country list, work-profile split-tunneling behavior, Private DNS interaction, and the update path for anything obtained outside the Play Store. Those are not footnotes. They are the difference between a tool that fits your setup and one that quietly does not, and the official channels at protonvpn.com are where the answers live.

Pros

  • Open-source Android client code
  • Swiss jurisdiction and public no-logs policy
  • Free tier available without ads

Cons

  • Free tier has limited server locations and speed
  • Some advanced features require a paid plan
  • Performance depends on network conditions and server load
Download from the official source

The link goes straight to Proton AG. We do not host app files.

Open official page →

Frequently asked questions

Is Proton VPN free?

The recorded pricing model is Free tier available; paid plans vary by region and billing period, exact prices unknown.

What license does it use?

It is listed under GPL-3.0-only.

Where can I download the official version?

The official link points to https://protonvpn.com/download.

This article was drafted with AI assistance using official developer documentation, release notes, privacy policies, and public data, then reviewed by an editor. We do not claim lab testing. The full method is on our Review Method page.