Internet Speed Test — Download, Upload, Latency & Jitter
A browser speed test that reports the four numbers that actually matter: download and upload throughput in Mbps, latency in milliseconds, and jitter — how much that latency wobbles, which is what makes calls stutter. Payload sizes adapt to your line, so a slow connection is not handed a 100 MB file and a gigabit connection still gets one big enough to saturate. Nothing runs until you press Start.
The one tool here that has to use the network: measurement traffic goes to Cloudflare's public speed endpoints (speed.cloudflare.com), and the uploaded bytes are random padding generated in your browser, never your files. orangebot.ai receives no request, stores no result, and keeps no history.
FreeNo sign-upRuns in your browserNetworkApplication
What each number needs to be
Speeds are quoted in megabits per second (Mbps); file sizes are in megabytes (MB), and there are 8 bits to a byte — so a 100 Mbps line downloads at about 12.5 MB/s at best. These are the practical thresholds behind the rating shown with your result.
| Activity | Download needed | Also needs |
|---|---|---|
| HD video call (1080p) | 3–4 Mbps | 3–4 Mbps upload, jitter under ~30 ms |
| 4K stream, one screen | 25 Mbps | Steady rate — buffering is usually jitter, not headline speed |
| Competitive gaming | 5–10 Mbps | Latency under ~50 ms matters far more than raw speed |
| Cloud backup / large uploads | n/a | Upload is the constraint; many cable plans give 10–40× less up than down |
| Household of four, mixed use | 100+ Mbps | Wi-Fi coverage, not the line, is the usual bottleneck at this level |
- Asymmetry is normal on cable and DSL: 500 Mbps down with 20 Mbps up is a common plan shape. Fibre is usually symmetric. If your upload figure looks tiny next to download, check your plan before assuming a fault.
- A single browser test is one sample. Congestion is time-of-day dependent, so a slow evening result and a fast morning result on the same line usually means contention in your neighbourhood, not a broken connection.
- This test uses one connection at a time. Some speed tests open many parallel streams, which can report a higher number on lossy links — neither is wrong, but it is why results differ between tools.
How to test your internet speed
- Free up the line. Pause big downloads, cloud backups, and 4K streams on other devices first. A speed test measures what is left over, so a busy network reports a slow line that is not actually slow.
- Press Start. No traffic is sent before you do. The test runs three phases in order: ten small round-trips for latency and jitter, then a download ramp, then an upload ramp — roughly 20 seconds in total.
- Read all four numbers. Download tells you how fast pages and video arrive; upload matters for video calls, backups, and sending files; latency drives how responsive everything feels; jitter above about 30 ms is what makes calls break up even on a fast line.
- Compare like for like. Run it two or three times, then repeat over Ethernet. If Wi-Fi is far below the wired result, the bottleneck is your Wi-Fi, not your ISP — which is the single most useful thing this test can tell you before you call support.
Frequently Asked Questions
- Where does the test traffic actually go?
- To Cloudflare. Downloads come from https://speed.cloudflare.com/__down and uploads go to https://speed.cloudflare.com/__up — the same public endpoints that power speed.cloudflare.com, served from whichever Cloudflare edge is closest to you. Those requests are subject to Cloudflare's own logging. orangebot.ai is not in the path at all: we receive no request, store no result, and keep no history, so reloading the page loses your run.
- What exactly do you upload?
- Random padding bytes generated in your browser with crypto.getRandomValues, nothing else. No file of yours is read, and there is no file picker on this page. The upload phase exists only to push bytes in the other direction so the return path can be timed.
- How much data does one test use?
- It scales with your connection: roughly 5–10 MB on a slow link and up to about 150 MB total on a fast one, because a fast line has to be given a big payload before it saturates. On a metered or mobile plan, keep that in mind before running it repeatedly.
- Why is my result lower than the speed I pay for?
- Several honest reasons stack up. Wi-Fi loses a large fraction of the wired rate, other devices share the line, HTTPS and TCP overhead cost a few percent, and ISPs advertise "up to" figures. A browser test measures what a browser can really pull, which is what matters for streaming and downloads. Retest over Ethernet with nothing else running before treating a gap as an ISP problem.
- What is jitter, and why does it matter more than it looks?
- Jitter is the standard deviation of the latency samples — how inconsistent your round-trip time is. A line with 20 ms latency and 5 ms jitter feels smooth; the same 20 ms latency with 60 ms jitter makes video calls stutter and games rubber-band, because packets arrive bunched and late. It is often the real culprit when a "fast" connection feels broken.
- Why does the download number exclude the first round-trip?
- Throughput is measured from the first byte received, not from when the request was sent. The gap before the first byte is latency, which already has its own phase — counting it twice would drag the Mbps figure down on high-latency connections for no good reason. Mbps itself is plain bytes × 8 ÷ seconds ÷ 1,000,000.
Use Cases
- Check whether you are getting the speed your ISP plan advertises
- Prove a Wi-Fi problem by comparing a wireless run against an Ethernet run
- Find the dead spots in a house by testing room by room
- Confirm upload headroom before a video call, live stream, or big backup
- Sanity-check a hotel, café, or co-working connection before committing to work on it