IP Speed Test · Proxy Node Bandwidth Test
IPOK Speed Test measures your proxy or network exit's download and upload bandwidth and latency across 9 global regions via Cloudflare Edge and M-Lab MSAK, quantifying real node performance.
One click measures your current download / upload bandwidth and round-trip latency to 9 global regions. With a proxy or VPN active, you're measuring the real speed of that exit node — run it before picking a node, switching lines, or renewing a plan.
Dual-source cross-validation: bandwidth is measured against both the Cloudflare edge network and an M-Lab server (MSAK protocol — the same backend behind Google's “speed test” widget). Two fully independent infrastructures — a QoS-whitelisted “inflated score” on one of them shows up immediately.
Results are pinned to your current exit IP — speed and IP inspection live in the same tool: after testing, review this exit's purity, risk score, and known streaming-region restrictions. IPOK does not run a live unlock test.
Speed test
Tests YOUR current connection (through your proxy if one is active): dual-source bandwidth = Cloudflare edge + M-Lab (MSAK); latency = browser round-trip to regional AWS endpoints. Runs only on click. IPOK does not write results to its application database; Cloudflare / AWS process network requests, and M-Lab records tests including IP and speed in its public dataset.
Click “Start network speed test” — dual-source download / upload bandwidth + 9-region global latency (~40s)
Reading a proxy or VPN node speed test
The download and upload rows show how much bandwidth is left after traffic passes through the node. With a node active the bottleneck is almost always the node, not your home line, so numbers below your broadband plan are normal. The useful comparison is relative: same device, same time, node A and then node B.
The latency table lists round trips from your browser, through the current exit, to endpoints in 9 regions. Read the row for the region your target service lives in: US West or US East for US streaming, the target market's region for cross-border commerce. The latency shown inside a proxy client is usually a single hop from your device to the node entry, which is a different measurement.
A wide gap between the two bandwidth sources is itself a clue. Cloudflare edge reflects ordinary CDN-backed browsing; M-Lab is closer to the raw capacity of the path. When one is much lower, the node is likely throttling or shaping traffic in that direction.
Speed and purity, side by side
Fast is not the same as clean. The exit IP pinned on the speed card is the one to run through the multi-source check on the home page: hosting or residential, flagged or not, and how the /24 neighborhood looks. When picking a node, look at both cards together. The one with usable speed and an acceptable purity result is the one worth keeping.
A speed test describes only this connection right now; switching nodes or testing at another hour changes the result. To keep evidence, note the exit IP and both bandwidth figures, then test again a week later at the same hour.
Check this exit's purity on the home page →
FAQ
What exactly is being measured?
Your browser's actual network path right now. With a proxy / VPN active, you measure the throughput after traffic passes through that node; without one, it's your local broadband. Switch nodes and hit “Run again” to compare lines.
Why dual-source? Isn't one speed test site enough?
Some ISPs and datacenters QoS-optimize traffic to well-known speed test servers, producing flattering but misleading numbers. Cloudflare edge and M-Lab are independent infrastructures — if the two figures roughly agree, trust them; a large gap means your line gets “special treatment” somewhere.
How is this different from Speedtest.net?
Speedtest measures peak throughput to ISP-curated servers, which are often specifically optimized. We measure your speed to generic infrastructure (Cloudflare edge / M-Lab), much closer to what you experience on ordinary websites. This page loads no third-party ad or ad-tracking scripts but may show clearly labeled affiliate recommendations; it uses no tracking cookies, and the test runs only on click.
The two sources disagree a lot — which one do I trust?
The gap itself is information. The two sources traverse entirely different target networks: Cloudflare edge reflects the experience of browsing ordinary CDN-backed websites, while M-Lab reflects the raw capacity of your line. A large gap usually means one path gets special treatment somewhere — e.g. a proxy provider throttling CF-bound traffic, or optimizing speed-test routes. Read the row that matches your actual workload.
Why is my result lower than my broadband plan?
Common causes: an active proxy (the node, not your broadband, is the bottleneck), congested cross-border links at peak hours, or a single browser connection not saturating a fat pipe. Relative comparison matters more than absolute numbers — test node A vs node B back to back.
Does the test leak my data?
Upload tests send meaningless filler bytes with no business content; results render only in your browser and IPOK does not write them to its application database. Cloudflare / AWS process the requests, and every run also contacts M-Lab, which records IP and speed in its public research dataset. If you do not want to participate in M-Lab, do not start this test.
Why does the latency in my proxy client differ from the latency here?
Most clients (Clash, Shadowrocket and similar) measure one hop from your device to the node entry, or the node's response time to a fixed URL. This page measures the full round trip from your browser through the exit to regional endpoints. The two use different yardsticks, so a gap of tens of milliseconds is normal. Watch the trend rather than the absolute number.
How can I tell whether a node is being throttled?
Test several times in the same hour and compare with another node. If both bandwidth sources sit far below the plan's stated speed every time, the node itself is probably short on bandwidth. If only one source is low, traffic shaping in that direction is the more likely explanation. IPOK cannot see a provider's throttling policy; these are circumstantial signs only.