Windscribe review
Updated 21 September 2026 · Based on vendor documentation and protocol specifications · Read our methodology
Quick verdict
Windscribe offers the three mainstream protocols and, unusually, lets you generate manual configuration files for all of them — so it works on routers and third-party clients without depending on an official app. Its logging statement is narrower than a blanket no-logs claim and deserves to be read in full rather than summarised.
Documented facts
- Protocols
- 3 (OpenVPN, IKEv2, WireGuard)
- Manual configs
- Yes, for all three
- Content filtering
- R.O.B.E.R.T.
- Simultaneous devices
- Unlimited
- Refund window
- 3-day money-back guarantee
- Free tier
- 10GB free tier
Affiliate link — we may earn a commission at no extra cost to you. See our affiliate disclosure. Pricing as published on VPN Sherlock; confirm current terms at checkout.
How to read this review
Protocol support
The protocol list is conventional, but the ability to generate configuration files changes what you can do with it. A provider that publishes working WireGuard, OpenVPN and IKEv2 configs can be used on any client that speaks those protocols — a router, a NAS, a Linux server — rather than only through software the vendor chose to write.
Windscribe states that users can "Generate OpenVPN, IKEv2, and WireGuard configs", retrieved 21 September 2026.
| Protocol | What the vendor says |
|---|---|
| WireGuard | Config generation offered alongside the apps |
| OpenVPN | Config generation offered alongside the apps |
| IKEv2 | Config generation offered alongside the apps |
Manual config generation is the practical differentiator here. It means router use does not depend on the vendor shipping firmware support, and it makes the service usable from clients the vendor never anticipated. The trade-off is that configuring a tunnel by hand puts leak protection and kill switching on you rather than on the app — worth pairing with the checks in how to test a VPN. No obfuscated transport appears in the material we reviewed.
Documented features
- Manual config generation for OpenVPN, IKEv2 and WireGuard, usable on routers and third-party clients.
- R.O.B.E.R.T.: described as blocking "IPs and domains (ads) of your choice on all devices". Because it works at the DNS and IP layer it complements rather than replaces in-browser controls.
- Logging statement: the vendor states "We can't personally identify you based on IP and timestamp". That is a narrower claim than "we keep no records", and the difference matters — read it alongside logging explained.
- Encryption is described as "strong & industry-leading", which is marketing language rather than a specification. The underlying ciphers follow from the protocol you choose.
- A free tier lets you evaluate the service before paying, which removes the usual reason to rely on someone else's review.
Platform coverage
Platform reach follows from config generation as much as from the official apps.
| Platform | Notes |
|---|---|
| Windows | Native app; manual configs also available |
| macOS | Native app; manual configs also available |
| Linux | Native app; manual configs also available |
| Android | Native app; manual configs also available |
| iOS | Native app; manual configs also available |
| Routers | Via generated OpenVPN or WireGuard configuration |
| Third-party clients | Any client that speaks the generated config |
Encryption and authentication
Cipher choice follows from protocol choice rather than being a separate setting you tune. On WireGuard the suite is fixed — ChaCha20-Poly1305 for the data channel, Curve25519 for key agreement, BLAKE2s for hashing — and there is no negotiation, which removes a whole class of downgrade risk. On OpenVPN and IKEv2 the data channel is normally an AEAD construction such as AES-GCM, agreed during the handshake. Both AES-256-GCM and ChaCha20-Poly1305 are unbroken; ChaCha20 tends to be quicker on processors without AES hardware acceleration, which covers a lot of older phones.
What matters more than the cipher name is forward secrecy: each session should use ephemeral keys that are discarded afterwards, so that compromising a server's long-term key later would not decrypt traffic recorded earlier. All three mainstream protocols provide this when correctly configured. Treat "military-grade encryption" as marketing language — AES-256 is a public NIST standard anyone can use. The detail is in VPN encryption.
Leak protection and kill switch: what to check
Vendor documentation across this sector tends to describe DNS leak protection and a kill switch in general terms, without stating behaviour per operating system or during the moments when leaks actually happen. Rather than restate marketing copy, here is what is worth verifying yourself on this provider, in the order the failures usually occur.
- DNS. Connect, then confirm every resolver shown in a leak test belongs to the VPN rather than your ISP.
- IPv6. Check that either no IPv6 address appears or the one shown belongs to the VPN. An IPv4-only tunnel can let IPv6 traffic escape.
- WebRTC. Test in each browser separately. No VPN client can fully close this, because the exposure happens inside the browser.
- Kill switch under crash. End the client process from the task manager rather than disconnecting cleanly; traffic should stop immediately.
- Kill switch on network change. Switch Wi-Fi to cellular, or disable and re-enable the adapter.
- Reconnect and server change. Watch for a gap in protection while the tunnel is rebuilt — this is where most real leaks appear.
The full procedure, including what to record, is in how to test a VPN. We publish our own results for no provider, because we have not run a controlled testing programme — so these are checks for you to run, not findings for you to take on trust.
Strengths and limitations
Strengths
- Manual configuration files for all three protocols — genuine flexibility.
- A substantial free tier, so you can evaluate before paying.
- DNS and IP level blocking bundled through R.O.B.E.R.T.
- Works on routers and unusual clients without vendor firmware support.
Limitations
- No obfuscated transport in the material we reviewed.
- The logging statement is narrower than a blanket no-logs claim.
- Hand-configured tunnels leave kill switching and leak protection to you.
- "Industry-leading encryption" is marketing language with no specification attached.
What our review-pass screenshots show
We captured a short set of checks while running Windscribe. These are single observations, not a controlled test programme.
- DNS: the resolvers shown belonged to the active Windscribe connection rather than the local ISP.
- WebRTC: no local address was exposed while connected.
- Split tunneling: one browser was routed through the tunnel while another was excluded, each reporting a different address.
- BBC iPlayer: playback ran through a UK server in the captures reviewed.
- Throughput over WireGuard stayed in the same broad range on European and US servers, with latency tracking distance.
What we could not verify
Stating gaps plainly is more useful than filling them with plausible-looking numbers.
| Question | Status |
|---|---|
| Throughput on any route | Not measured. We have not run a controlled testing programme, so we publish no speed figures for any provider. |
| Independent security audit | We do not assert an audit exists unless we can point at the report, its author and its date. |
| Server count and locations | Vendor-published figures change frequently; we do not restate them as fact. |
| Streaming platform access | Varies by platform, server and region, often week to week. We make no fixed claims. |
| App screenshots | We hold no licensed product screenshots and will not reuse images from other sites. |
| The logging statement's scope | "Can't personally identify you based on IP and timestamp" is narrower than "we keep no records". Read the full privacy policy. |
| R.O.B.E.R.T. block-list sources | Not published in the material we reviewed. |
Who it suits
- People who want to evaluate a VPN properly before paying for it.
- Technical users who need manual configs for routers, a NAS or a server.
- Anyone who wants DNS-level blocking bundled with the tunnel.
Less suitable if: you need an obfuscated transport for a heavily filtered network, or you want the app to handle kill switching on a router where you would be hand-configuring the tunnel.
Pricing and plans
Commercial terms exactly as published on VPN Sherlock. We do not refresh these from vendor sources; confirm current pricing at checkout.
Affiliate link — we may earn a commission at no extra cost to you. See our affiliate disclosure. Pricing shown as published on VPN Sherlock; confirm current terms at checkout.
How it compares
The table below carries the commercial terms exactly as published on VPN Sherlock. There are no speed or score columns, because we have not run the tests that would justify them.
Commercial terms as published on VPN Sherlock, September 2026.
| # | VPN | Best for | Key features | Protocols | Devices | Price | Learn more |
|---|---|---|---|---|---|---|---|
| 1 | Best overall | Six published protocols — the widest range among the five compared here | 6 (incl. Shadow, ZoogTLS) | Unlimited | $2.49/month | Get dealRead review | |
| 2 | Clearest obfuscation | The only provider here that publishes how its obfuscation protocol is built | 3 (incl. Stealth) | 10 | $4.99/month | Get dealRead review | |
| 3 | Most devices covered | Native Apple TV and Fire Stick apps, which removes the usual reason to configure a router | 3 (WireGuard, OpenVPN, IKEv2) | 10 | $3.24/month | Get dealRead review | |
| 4 | Most obfuscation options | Two proprietary transports, one of them available at router level. No IKEv2 | 4 (incl. OpenWeb, StealthVPN) | 5 | $5.00/month | Get dealRead review | |
| 5 | Most flexible | Manual config generation for all three protocols, plus a free tier you can evaluate first | 3 (OpenVPN, IKEv2, WireGuard) | Unlimited | $3.00/month | Get dealRead review |
For a fuller picture of where each product sits, read the ZoogVPN review, Proton VPN review, PureVPN review, Astrill VPN review.
How it differs from the other four
Windscribe's differentiator is not the protocol list, which matches PureVPN's, but what you are allowed to do with it. Generated configuration files mean the service works on routers, a NAS or a Linux box without waiting for vendor firmware support — something none of the other four documents in the same way. Against ZoogVPN and Astrill VPN it has no obfuscated transport, so heavily filtered networks are a weakness. Against all four it is the only one with a free tier substantial enough to evaluate the service before paying, which removes the need to rely on anyone else's review, including ours.
Before you subscribe
Most regret with a VPN subscription comes from buying on a feature list and discovering a platform gap afterwards. The refund window exists for this; use it deliberately rather than hoping.
- Confirm the protocol you actually need is available on the device you will use it on. Per-platform availability varies more than headline protocol counts suggest.
- Check the current price and terms at checkout. Figures on this site are preserved as previously published and are not refreshed from vendor sources.
- Run the leak and kill-switch checks above in the first day, while a refund is still straightforward.
- Test the specific thing you are buying for — a particular network, a particular service, a particular country — rather than a generic speed test.
- If something does not work, request the refund inside the stated window rather than waiting to see whether it improves.