Internet Speed Test
Download, upload, ping, jitter and latency under load — measured against RahaKit’s own Worker endpoints, with plain-language scores for streaming, gaming and video calls.
Want the address behind the numbers? See your full IP report
4K needs 25
Calls need 5
Gaming wants <50
Calls want <30
Each bar fills to where the value lands on its scale — the bold word is the verdict, and the note beside it says what the activity needs.
Answered by…
Data transferred0 MB
Runs in this tab—
Best download—
Median download—
Latest—
Client…
Server…
Address family…
ASN…
AS organisation…
Non-linear scale — reads slow lines and gigabit lines on one dial. Run history lives in this tab’s memory only — nothing is kept after this tab closes.
Press Start test when you are ready. Nothing is measured, and no data is exchanged for measuring, before that.
Measurement details
What this connection is good for
Quality scores from this run’s own measurements, using Cloudflare’s published AIM method.
Streaming
Not run yetGaming
Not run yetVideo calls
Not run yetPacket loss is not measured in this version of the test, so the scores are calculated without it and may read lower than a full test. Scores are estimates from a short test — your results can vary by time of day.
Run the test
Your browser exchanges timed payloads with our own Worker endpoints — the same open-source measurement engine as Cloudflare’s own speed test, with its result reporting switched off. A full run uses up to about 190 MB on a fast line.
Your result
When a run finishes, the final values sit in the grid above: download and upload in megabits per second (Mbps), ping and jitter in milliseconds (ms), and the latency measured while each transfer was under way. Those numbers describe one specific route — the connection between this device and RahaKit’s Worker infrastructure, which answers from the nearest Cloudflare data centre, shown by its code in the result. They are an honest measurement of that route, not a verdict on your whole internet connection and not your provider’s advertised rate. Download and upload are measured one transfer stream at a time — the way a single download or upload in your browser behaves — so tests that open many streams at once can read higher. Below the grid, the same measurements are translated into quality scores for streaming, gaming and video calls.
Download speed
Download speed is how fast data travels from the internet to your device — the direction that carries web pages, videos, music, app updates and file downloads. It is measured in megabits per second (Mbps): this test times real data moving from our Worker endpoints to your browser and divides the amount by the time it took. Most everyday activity is download-heavy, which is why this is usually the biggest number on a home connection.
Upload speed
Upload speed is the opposite direction: how fast your device can send data out — the photos and videos you share, the files you back up, the picture and sound you send on a video call. It is measured the same way, in Mbps, by timing data your browser sends to our Worker. On many home plans upload is deliberately smaller than download, so a lower upload figure is normal and not a fault.
Ping and jitter
Ping is the round-trip time for a tiny request: how long it takes a small fetch to reach the server and come back, in milliseconds. Lower is more responsive — it is the delay you feel between clicking and something happening, and the lag figure gamers watch. Jitter is how much that round-trip time varies from one ping to the next. A steady 40 ms feels smoother than a ping that jumps between 15 ms and 90 ms, even though the averages look similar.
Latency under load
Idle ping is measured while your line is quiet. Latency under load is measured while the line is busy — here, while the download or upload transfer is running at full pace. Many connections stay quick when idle but slow down noticeably when a big transfer fills the line, and that loaded figure is the one you feel when a call stutters during someone else’s download. This page reports latency under load separately for the download and the upload phase, alongside the idle ping, so you can compare the two states honestly.
What affects a result
Wi-Fi is the most common factor: distance from the router, walls, and neighbours’ networks all cost speed, and a cable connection will usually beat wireless on the same line. Other people and devices sharing the connection take their share while you test. Your own device matters too — an old phone, a busy laptop or a browser under load can cap the result before the line does. Finally, the route matters: the test talks to the Cloudflare data centre that answers it, and conditions on the path there change through the day.
What this test does not measure
The honest limits of the numbers above.
- It measures the path between this device and the Cloudflare data centre that answers your connection. That centre’s code is shown with your result — it is not a server you picked, and it is not a map of your route.
- There is no server choice: the same RahaKit Worker endpoints answer every run.
- Packet loss is not measured in this version: it needs a UDP relay step this test does not include. The quality scores are calculated without it, are labelled, and may read lower than a full test.
- Download and upload use one transfer stream at a time per direction. Tests that open many streams at once can report higher numbers on the same line.
- Ping is measured over HTTP, so it includes HTTP overhead and reads a little higher than a bare network ping.
- Very fast connections may be limited by the device or the browser before the line itself is the limit.
- Each run is a snapshot of one moment; conditions change, and another run may read differently.
- Run history (best, median and latest) is kept in this tab’s memory only, and is gone when the tab closes.
Privacy and data handling
The test runs entirely between your browser and RahaKit’s own Worker endpoints — there is no third-party speed-test service involved, and nothing is sent anywhere else. The payload data exchanged is generated filler whose only job is to be timed. Results, run history and data totals stay in this tab’s memory and disappear when the tab closes. Nothing is stored, nothing is logged, and there is no account. The measurement engine is open source; its own built-in result reporting — which would send results to Cloudflare — is switched off in our build, and a run’s network traffic goes only to rahakit.com. You can check that in your browser’s network tab.
Frequently asked questions
How does this speed test work?
When you press Start, your browser exchanges timed payloads with RahaKit’s own Worker endpoints: small fetches for ping and jitter, larger transfers for download and upload, with extra small fetches during the transfers for latency under load. Speeds are amounts divided by the times taken.
Why is my result different from other speed tests?
Different tests talk to different servers over different routes, use different payload sizes and methods, and run at different moments. A test measuring to a server in your city and one measuring to the nearest Cloudflare data centre are timing different journeys, so the numbers legitimately differ.
What is a good result?
It depends on the job. Video calls want a few Mbps each way and a steady, low ping. 4K streaming wants roughly 25 Mbps of download per screen. Gaming cares less about raw speed and more about low, stable ping and jitter. Big uploads — backups, video files — simply finish sooner with more upload Mbps.
Does running the test use much data?
Each run moves real payload to measure with, and faster lines move more of it in the same time. The page shows the data transferred for every run, so you can see the cost before repeating it. On a capped plan, avoid running it back to back.
Why is Wi-Fi slower than a cable?
Wi-Fi shares the air with walls, distance, interference and every other device on the same channel, and its speed falls as you move away from the router. A cable gives your device a dedicated path to the router, so it usually reaches the line’s true speed more reliably. Testing both ways shows you the gap.
Why does the ping here read higher than elsewhere?
Ping on this page is measured over HTTP, the same protocol web pages use, so it includes the overhead of setting up and answering an HTTP request. A bare network ping skips that work. The figure here is the delay a real web request experiences, which is why it can read a little higher.
Are my results saved anywhere?
No. Results and the run history (best, median and latest) live only in this tab’s memory and are erased when the tab closes. Nothing is stored on our side, nothing is logged, and there is no account to save them to.
Does it work on a mobile connection?
Yes. The test runs in the browser, so it works on mobile data and Wi-Fi alike. Mobile results vary more — signal strength, network generation and movement all change mid-test — and each run uses some of your data allowance, which the page counts for you.
What is latency under load?
It is how much your ping rises while the line is busy moving a big transfer. This test samples ping during the download and during the upload and reports both, plus the increase over your idle ping. A small increase means the connection stays responsive while it works; a large one is why calls stutter when someone else starts a big download.
What do the quality scores mean?
Streaming, gaming and video-call scores translate the measured numbers into a plain label — Bad, Poor, Average, Good or Great — using Cloudflare’s published AIM method. They are estimates from this run’s measurements, not a test of any specific app or service, and they always sit next to the numbers they came from.
Why is packet loss not measured?
Packet loss is measured with UDP packets, and this version of the test does not include the UDP relay step. Because packet loss is one of the AIM inputs, the quality scores are calculated without it and are labelled that they may read lower than a full test. We would rather show “not measured” than estimate a number we did not measure.
What is a good ping for gaming?
Under about 30 ms to the test endpoint feels responsive for most games; 30–60 ms is playable for many; above 100 ms most players notice the delay. Your ping to a specific game server will differ — it depends on where that server is and the route to it.
Back to the whole picture: See your full IP report — your addresses, location, network and connection, in one place.
The other half of the picture: See your request headers — the lines your browser sends to every site it visits.