Account

What new tool do you want?

Developer Tools

Internet Speed Test

Test your internet speed — download, upload, ping and jitter — in about 15 seconds, with a plain-English verdict on what your connection can handle.

How to use

  1. 1Press Start test.
  2. 2Wait about 15 seconds while download and upload are measured.
  3. 3Read your results and what they mean for streaming, calls and gaming.

Frequently asked questions

Why is my result lower than my plan?

Wi-Fi signal, other devices using the network, VPNs and the time of day all reduce speed. For the truest number, test on a cable connection with nothing else running.

What is a good ping?

Under 20 ms is excellent, under 50 ms is good for gaming and calls; above 100 ms you may notice lag.

How much data does it use?

Up to about 190 MB on very fast connections, much less on slower ones.

What's being measured, and against what

This test talks to Cloudflare's public speed-test endpoints (speed.cloudflare.com), the same infrastructure Cloudflare's own tool uses — so results reflect the path between your browser and Cloudflare's nearest data center, not a lab-only loopback. Latency is sampled first: 12 tiny requests are fired in sequence, the first two are discarded as warm-up, and the result is the median of the remaining ten, with jitter computed as the average absolute difference between consecutive samples. Download runs four parallel connections for up to 6 seconds (capped at 150 MB so a very fast line doesn't burn unlimited data); upload runs three connections for up to 6 seconds, capped at 40 MB.

Why your number won't match your plan

A plan's advertised speed is the maximum the ISP's line can deliver, not what the browser will measure at any given moment. Wi-Fi contention, other devices on the network, VPNs that add an encryption and routing hop, and even the time of day (ISP congestion is worst during evening peak hours) all pull the measured number below the plan figure. For a number close to the line's real capacity, test over Ethernet with other devices idle.

Worked example: reading the verdict

Say a test finishes with 32 Mbps down, 9 Mbps up, 18 ms ping and 2.1 ms jitter. The tool's rating logic checks download against fixed thresholds — 100+ Mbps is rated great for 4K on several devices, 25–99 good for HD/4K streaming and gaming, 10–24 okay for HD on one or two devices, under 10 slow. 32 Mbps clears the 25 Mbps bar, so the verdict shown is 'Good for HD/4K streaming, gaming and video calls' — even though it's far below a typical 500 Mbps plan, because the rating is about what the connection can actually do, not how close it came to the plan's ceiling.

Why jitter matters even when download looks fine

Jitter measures how much consecutive ping samples vary from each other, not how high ping is on average — a connection with a high but rock-steady 80ms ping can feel smoother for a video call than one averaging 20ms but swinging wildly between 5ms and 60ms, because real-time audio and video depend on packets arriving at a predictable, consistent interval, not just a low average delay. That's why this test reports jitter as a separate number from ping rather than folding it into a single combined 'latency' score: a good download speed and low average ping can still mask an unstable, jitter-heavy connection that stutters on calls.

Why the test uses multiple parallel connections instead of one

Download runs four simultaneous connections (upload uses three) rather than measuring a single stream, because a single TCP connection often can't saturate a fast line on its own — TCP's congestion-control algorithms ramp up a single connection's throughput gradually, and on very fast connections or those with higher latency to the test server, one stream may plateau well below the line's real capacity before the test window ends. Multiple parallel connections more accurately reflect how real usage looks anyway, since loading a modern webpage or streaming video typically opens several connections at once rather than one.

Related tools

More: Developer Tools