There is no single answer to whether a free VPN or a paid one is better, but there is a repeatable order for deciding: first lay out the four hidden costs of anything free — throttling, data allowance, privacy and ads — then compare them with how often and for what you actually use it. This article follows that order, marking where free is good enough, the point at which paying wins, and the things to confirm before you move from free to paid.
The short version first: for occasional, short sessions where stability doesn't matter, a free plan holds up. Once cross-border access becomes part of your daily workflow, the time and attention a free plan eats usually cost more than a monthly subscription.
Start by counting the four costs of free
Running a free service costs real money: egress bandwidth is billed by traffic, IP ranges are rented by the block, and the client has to be maintained. With no subscription revenue, that money has to come back from somewhere, and it usually comes from four places — the four rows in the table below.
| Cost | What free services usually do | Paid plans (this site as the example) | What it means for you |
|---|---|---|---|
| Bandwidth | Per-user egress is squeezed at peak hours, and many users share the same nodes | Monthly plans: ¥9.9 for 60GB, ¥18 for 250GB, ¥28 for 500GB | Video calls drop frames and pages keep spinning |
| Data allowance | A small daily or monthly quota; once it runs out you are throttled or disconnected | Data resets every month on your activation date, and mid-cycle upgrades are prorated against the remaining days | When the quota runs dry, the work doesn't get finished |
| Privacy | Costs are subsidised with ads, referral traffic and data monetisation, and the logging scope is opaque | No email address needed — a username and password are enough | You can't tell where browsing history and device details end up |
| Devices | Simultaneous connections are capped and sharing with family is not allowed | No device limit | Devices have to take turns, knocking each other offline |
The table has no column priced at zero, because free was never free — it just moves the cost from your bill to your time, your quota and your uncertainty. The real comparison isn't ¥0 versus ¥9.9; it's whether the money you save is worth more than the time you spend.
Throttling and data caps: the two hardest walls in free plans
Bandwidth is the first thing free plans cut. Egress is billed at peak rates, so the more users there are, the less each one gets — which is why free services generally do two things at peak hours: cap the speed of each connection and cut the number of usable nodes. The result is that pages still open in the morning, video starts buffering after 8 p.m., and switching nodes only takes you from slow to slower.
The data allowance is the second wall. The usual pattern is a small daily or monthly quota, after which you are either disconnected or throttled to the point where images won't load. To work out whether you'll hit it, divide the quota by your real usage: on the 60GB tier, two hours of HD video a day plus meetings and cloud sync will often run dry before the month ends; if you only look things up and answer email, the same quota goes a long way.
One more limit is easy to miss: concurrent connections and session length. Free clients often disconnect and reconnect on a timer in the background, labelled "auto-optimise" in the interface when the real purpose is to recycle node capacity. Web browsing barely notices, but for SSH sessions, long uploads or online meetings — anything that needs a persistent connection — one drop means starting over.
Privacy and ads: how free services stay alive
This one deserves more suspicion than speed, because it doesn't show up immediately. Free clients typically monetise in three ways: ads or push notifications inside the app; funneling new users to landing pages; and bundling connection logs and device details for third parties. The first two you can see. The third usually sits in the privacy policy, buried in vague wording.
A few concrete things tell you whether a free client is worth using:
- Whether the permissions it requests at install have anything to do with tunnelling. A client that only forwards network traffic has no business reading your app list or file directories.
- Whether the privacy policy states which fields are logged, how long they are kept and whether they are shared with third parties. A sentence like "we may collect necessary information" says nothing at all.
- Whether it offers DNS handling and split-tunnelling settings. Those two decide whether your browsing records leak.
DNS leaks: the most overlooked link
Once the tunnel is up, IP-layer traffic really does go through the proxy — but domain resolution may not. If the client doesn't take over DNS, the system still queries your local ISP, which means the content is encrypted while the domains you visit are in plain text. At best, resolution is poisoned and you land on the wrong address; at worst, a complete record of your visits stays on the local network.
Testing it isn't complicated: connect, open any DNS leak test page, and check whether the resolver's location matches your exit region. If it doesn't, DNS is still local. Paid clients usually offer a "DNS handled by the tunnel" switch, or let you point DNS at a resolver inside the tunnel; free clients often only have a global mode with no split rules at all, so your local banking and corporate intranet get routed through the tunnel too — which makes risk controls more likely to trigger, not less.
The subscription link itself carries identity: anyone who has it can pull the same node list and burn the same quota. The "node sharing" popular in free tools is the same risk as pasting your subscription link into a public group. When you switch providers, delete the old subscription from the client first, then import the new one.
When free is enough
Turn those costs into concrete situations and the boundary gets much clearer. The list below maps directly onto how you use it:
- ✅ A quick lookup: you're out and need to open a few overseas sites for documentation, ten minutes at a time, then close it.
- ✅ A one-off check: confirming whether a service works on another network, for a few minutes only.
- ✅ Small downloads: an installer of a few tens of MB, an image, a PDF.
- ❌ Long video meetings: jitter on free nodes breaks up the audio, and one reconnect costs you half a sentence.
- ❌ Large transfers and cloud sync: the quota runs out within days, and after that you're disconnected or throttled.
- ❌ Admin panels that need a stable login: the exit IP keeps changing, so account security keeps demanding a second factor.
- ❌ Work material: a service with unclear logging has no business carrying contracts, code or customer data.
When paid is the better deal
What you buy isn't "faster internet" but predictable resources: a stated quota, stated reset rules, a stated refund window, and a support ticket when something goes wrong. The test comes down to one question — is cross-border access part of your workflow, or an occasional action?
If it's the former, a simple time calculation settles it: the minutes lost each day to reconnecting, throttling and switching nodes usually add up to more than a monthly subscription. Monthly plans here start at ¥9.9 for 60GB; if your usage is mostly video and transfers, the ¥18 250GB or ¥28 500GB tier fits better. Data resets monthly on your activation date, and mid-cycle upgrades are prorated against the remaining days, so there's no need to time your purchase to the start of a month or wait for a reset.
If your usage is uneven, look at the data packs: ¥158/300GB, ¥358/1000GB and ¥658/3000GB, valid until used up with no expiry — a good fit for people who travel heavily some months and barely use it in others. They're a different pricing model from the monthly plans; pick whichever matches your rhythm.
Moving from free to paid: five things to confirm first
Once you've decided to pay, don't rush the order. Confirming the five points below saves most of the "bought it and it doesn't fit" rework.
- Measure your monthly usage first. Check a week of real traffic in the old client, multiply by four, then add 30% headroom — that tells you whether you land on 60GB, 250GB or 500GB.
- Read the reset and upgrade rules. Data resets monthly on your activation date, and mid-cycle upgrades are prorated against the remaining days — which means you can pick your own start date instead of fitting around the calendar month.
- Confirm the refund window and payment methods. A full no-questions refund can be requested within 60 days of your first payment, which covers a complete billing cycle; payment supports Alipay, WeChat Pay and USDT, and Alipay receipts are the simplest route if you need to claim it as an expense.
- Confirm devices and platforms. There's no device limit, and Windows / macOS / iOS / Android / Linux all have clients, so several devices can stay online at once without knocking each other off.
- Migrate the subscription and clean up the old client. Paid nodes are delivered through a subscription link, and the client pulls updates on a schedule; quit the old free client and turn off its background proxy first, otherwise the two fight over the routing table and you get the phantom fault of "connected but no pages load".
Two things to do right after importing the subscription
First, look at handshake time, not bandwidth. The latency shown in the client is the round-trip time to the node: it tells you whether the route detours, but it isn't download speed. Use it to filter out obviously indirect routes, then verify with real page loads and a meeting.
Second, set up split-tunnelling rules. The usual approach is direct for local sites and proxied for overseas ones, with a dedicated route for services that need a stable long connection, such as meetings and code hosting. On route types: IEPL dedicated lines run on a separate channel and show the least jitter at peak hours; relayed routes add a hop and cost less; direct routes are cheapest and suffer most from congestion at the international gateway. A single subscription usually includes all three — just assign them by purpose.
FAQ
Do free VPNs throttle your speed?
Yes, and usually by time of day. Free services pay for bandwidth at peak rates, so the more users they have, the more they have to throttle at peak — which is why a node can feel fine at 10 a.m. and be unusable at 9 p.m. The throttling isn't a fault; it's the cost structure.
Is a paid VPN always faster?
No. Speed depends on three variables: your local network, the route type and the time of day. What you pay for is resource allocation and predictability — quota, reset rules and refund window stated up front, and a support ticket when something breaks. If your local connection to the international gateway is already congested, no provider will change that.
Do I need to pay if I only browse the web?
No. For looking things up, reading documentation and the occasional overseas site, the time cost of a free plan is acceptable. The real signal to pay is when you start postponing work because the connection is unstable, or when you need to stay connected for long stretches.
What if I want to stop after a while?
A full no-questions refund can be requested within 60 days of your first payment; the terms page has the full wording. Past that window you can cancel the monthly plan and cover the remaining need with a data pack — packs are valid until used up with no expiry, so nothing is wasted when you stop.