Internet Speed Test — Download, Upload, Ping and Jitter
This internet speed test measures your download speed, upload speed, ping and jitter in about fifteen seconds, straight from the browser — no app, no plugin, no account. It works the same as a WiFi speed test on a phone or laptop: it measures whatever connection the device is using right now. Most people land here because a result came back lower than expected, and the cause is rarely the one they assumed.
How this internet speed test measures your connection
The test runs three stages on the gauge: ping, then download, then upload. All three run against Cloudflare's public speed test endpoints, which answer from the Cloudflare data centre nearest to you — so the figures describe the path from your device to a nearby, well-connected server, not to a distant one on another continent.
Download streams real payloads of 100 KB, 1 MB and 10 MB. The first 0.7 seconds are thrown away, because a fresh connection starts slowly while TCP ramps up, and the reported figure is the average over the next five seconds. Upload sends incompressible random data in rounds that double in size, from 256 KB up to 8 MB, for up to six seconds, and only bytes the server confirms it received are counted. If a stage fails, the panel says so instead of showing a number nothing produced.
Two details matter more than they look. Loaded ping is sampled while the download and upload are saturating the link — it explains a call breaking up the moment someone starts a large download. And the Multi/Single switch is real: Multi runs five parallel connections, Single runs one, and the gap between them is worth running twice to see. Above the gauge the page also shows your public IP address, ISP and ASN, so you can confirm which connection — and whether a VPN — you are actually testing.
Speed test results explained: download, upload, ping and jitter
Four numbers, four different questions — and they fail independently. A fast download on a connection that feels broken is a normal combination.
Download speed (Mbps)
How fast data arrives. Mbps means megabits per second, not megabytes — the unit that trips up more people than anything else here. Divide by eight for the MB/s your download manager shows: a 100 Mbps line tops out near 12.5 MB/s, so 6 MB/s on a 50 Mbps plan is not a fault but the same speed in different units.
Upload speed (Mbps)
How fast data leaves. On fibre it is often symmetrical; on cable, DSL and 4G/5G it is frequently a tenth of download, because those technologies hand most of the channel downstream. Video calls, cloud backup and sending large files depend on it, so a 900/40 line can feel worse in a meeting than a 200/200 one.
Ping (ms)
How long a round trip takes, reported as the best of ten samples. It depends on distance and routing, not bandwidth — a faster plan does not lower it. Ping decides whether a game, call or remote desktop feels responsive. The ping test focuses on it, with the ranges that matter for gaming and calls and what to do about spikes.
Jitter (ms)
How much ping changes from one round trip to the next. A steady 60 ms is comfortable; a link swinging between 20 ms and 200 ms can average out respectably and still sound terrible on a call, because voice and video buffer for the worst case. High jitter points to congestion, WiFi interference, or a router queueing badly under load.
Why your internet speed is lower than the plan you pay for
A plan speed is an up to figure, measured where the ISP hands the line over: wired, into their box, with nothing else running. Everything between that point and this browser tab can only subtract.
WiFi is usually the bottleneck, not the line
WiFi is shared airtime, not a private cable. Devices take turns, walls weaken the signal, and neighbours compete for the same channels. A 2.4 GHz connection realistically delivers a few tens of Mbps whatever the plan says, and 5 GHz drops off quickly through walls.
Peak hours
Cable, fixed wireless and 5G home internet share neighbourhood capacity, so from 7pm to 11pm you compete with everyone around you; fibre-to-the-home is far less affected. If the morning test shows full speed and the evening test shows a third of it, that is your answer — check at 10am and 9pm on three weekdays before calling anyone.
The router or ISP gateway
Old hardware quietly caps everything behind it. Plenty of ISP boxes and older switches still have 100 Mbps Fast Ethernet ports, which pin a gigabit plan near 94 Mbps however you test. The giveaway is a result landing on exactly the same ceiling every run.
The device you are testing on
A cheap phone or a ten-year-old laptop runs out of CPU or radio capacity long before the line does. Background sync, a VPN, a torrent client, or the network-prioritisation software bundled with some gaming motherboards can throttle one machine while every other device tests fine. If one device is slow and the rest are not, the line is not the problem.
WiFi speed test vs wired: find where the speed is going
One test tells you little; two isolate the fault in five minutes. Plug a laptop into the router with an Ethernet cable and run the test, then unplug it, stay in the same room and run a WiFi speed test on the same laptop.
If wired is close to your plan and WiFi is half of it, the problem is inside your home — router placement, band, channel congestion or the router's radio — and no complaint to your ISP will fix it. If wired is also low, the problem is upstream and you have something worth reporting.
Then test WiFi where you actually sit: the desk, the bedroom, the far end of the house. Note the download, upload and loaded ping in each spot. A room where download halves and loaded ping jumps is the room that justifies moving the router or adding a mesh node — and checking the same spots again after the change shows whether it worked.
How to test internet speed accurately
- Stop other traffic first. Pause downloads, game updates, cloud backup and 4K streams on every device — one console updating in the background can halve your result.
- Close other tabs. The test competes with whatever else the browser is doing.
- Turn the VPN off. It adds a hop, encryption overhead and its own capacity limit. Test without it, then with it, to see the cost — the IP and ISP shown above the gauge tell you which one you are on.
- Run three times and take the middle result. Any single run can be 10–20% off because of routing and server load.
- Change one thing at a time. Same device, spot and hour — otherwise you are comparing two different experiments.
What a speed test cannot tell you about your internet
Worth being blunt, because plenty of arguments with ISPs start here. This measures this browser's throughput to one nearby server, at this moment, over the link you happen to be on. It is not a measurement of your ISP plan. It is the speed you got once, along one path: useful for diagnosis, close to useless as proof of a contract breach.
Three things cap the result before your ISP does. Connection count: a single stream is limited by TCP behaviour, latency and packet loss, and reports well below what several parallel streams reach on the same line — which is why tools that open many connections publish the biggest numbers. The device: on a budget phone or older laptop, the CPU handling encryption and browser overhead tops out first, and you end up measuring hardware instead of the link. WiFi: on wireless you are measuring the weakest wireless hop.
Ping here is the round trip of an HTTP request rather than an ICMP echo, so it sits a little above what the ping command reports. On multi-gigabit lines a browser test also tends to read below the real line rate, because the browser itself becomes the limit.
It says nothing about stability over time, either. A clean run proves the path was fine for a few seconds; it does not rule out the packet loss, latency spikes and bufferbloat that ruin games and calls while throughput looks perfect. Running the ping test repeatedly, idle and again under load, is how you find that. For a formal complaint, most ISPs only accept results taken wired into their own equipment.
Frequently asked questions
Why is my internet significantly slower than what I'm paying for?
Something between the ISP's handover point and your device is capping it — most often Wi-Fi, an old gateway, or neighbourhood congestion. Test wired into the router first. If wired matches your plan, the loss is inside your home. If wired is also far below it, check that your account is actually provisioned at the tier you pay for; billing and provisioning mismatches after a move or an upgrade are common and invisible from your end.
Why are my download speeds over WiFi 10-20x slower than Ethernet?
Your wireless hop, not your internet line, is the limit. A gap that large usually means you are on the 2.4 GHz band, on an ISP-supplied router with a weak radio, or too far from it through walls. Move to 5 GHz, test in the same room as the router, and if the gap stays, put a decent router or mesh node behind the ISP box and turn that box's Wi-Fi off.
My speed test says 50-60 Mbps but when I download something it never goes above 6 MB/s. Why?
Those are the same speed in different units. Speed tests report megabits per second (Mbps); download managers report megabytes per second (MB/s), and there are eight bits in a byte. 50–60 Mbps divided by eight is 6.25–7.5 MB/s, so 6 MB/s is exactly what your connection should deliver. Nothing is broken — ISPs advertise in bits because the number looks eight times bigger.
Why do I get completely different results on different speed test sites?
Because each site tests a different server over a different route with a different method. Server distance, how loaded that server is, how many parallel connections the tool opens, and how your ISP treats that particular traffic all move the number. A result that is wildly low on one site but normal on two others points at that test server, not your line. Compare a site against itself over time instead.
How much internet speed should I actually get?
For most households 100 Mbps download is comfortable, and anything above 300 Mbps is bought for peace of mind rather than need. A 4K stream uses roughly 25 Mbps, a video call 3–5 Mbps, and online gaming under 1 Mbps — gaming needs low ping, not bandwidth. Upload matters more than people expect: if you back up to the cloud or stream, look for 20 Mbps or better upstream.
My speed test is full speed but browsing is slow in the evenings. What's causing that?
Latency, not bandwidth. A speed test measures how much data fits through the pipe; browsing feel depends on how quickly each round trip completes, and evening congestion inflates round-trip time and jitter long before it dents throughput. Round trips jumping to 150–200 ms while your local network stays in single digits is the signature. Run the ping test a few times during the slow window — that is what shows the spikes.
Why is my upload speed so slow when my download speed is fine?
Either your plan is asymmetric, or something on that one device is throttling it. Cable, DSL and mobile plans normally give upload a fraction of download, so 500/20 is by design. But if other devices upload fine and one machine sits at about 1 Mbps, suspect local software: the network-prioritisation utilities shipped with gaming motherboards and network cards are a repeat offender, as are stale drivers and a damaged Ethernet cable.