iSpeedTestWhat the four measurements tell you
A speed test is an observation of the path between your browser and a measurement endpoint. Download describes incoming data, upload describes outgoing data, and HTTP latency describes how long a small request takes to finish. Jitter summarizes the change between consecutive idle timing samples. A high download figure cannot compensate for unstable timing during a call.
The tool above measures real transfers after you select Start. It uses Cloudflare HTTP endpoints, generated payloads and a limited data budget. It does not identify your provider, certify a plan, or measure every destination on the internet. Results are useful when they answer a specific question about how your connection behaves.
Get a repeatable baseline
Pause large transfers, keep the tab visible and use the device on which you experience the problem. If possible, compare with Ethernet using a capable cable and adapter. Record whether a VPN is active. Keep the same mode and data profile when comparing runs so a configuration change is not mistaken for a network change.
A full standard run allows up to 96 MB of test payload; Data saver allows up to 12 MB. Actual use can be lower, and protocol overhead is additional. There is no automatic bandwidth test on page load. Stop cancels outstanding work, although bytes already transferred still count toward your allowance.
Interpret a result in the context of a task
Moving a large file needs sustained throughput. A meeting needs sufficient capacity in both directions and consistent delivery. An online game can be sensitive to delay even when updates download quickly. Use the good-speed guide to plan concurrent activities, the file calculators to estimate transfer time, and the timing tools to inspect response variation.
Different services test different destinations, workloads and durations. A short bounded request may under-read a very fast line. A remote application can also be slower than the measurement endpoint. Repeat a controlled experiment before treating one result as a reason to buy a new plan or replace hardware.
Choose a diagnostic that answers your next question
Use the Wi-Fi comparison to record two rooms or a wired and wireless baseline on the same device. The mobile page begins with Data saver and helps label cellular, Wi-Fi and hotspot conditions. The broadband page records your subscribed rates beside completed measurements, without treating the difference as a contractual finding.
For slow uploads or downloads, estimate file time and check whether the application, server or storage is the actual limit. For delayed interactions, inspect the idle samples and then the loaded-latency observations. The troubleshooting guide provides a sequence for separating one-device, wireless, destination and shared-network problems.
Understand what stays on your device
Optional history keeps up to ten completed measurements in this browser. It is off until you enable it. Comparison labels and the current session record remain in the open page only; leaving or reloading clears them. Neither is a public score or an account-synced backup.
Cloudflare receives the measurement requests and their connection information. The site also uses Google Analytics for page and categorical feature usage. Room labels, plan entries and raw measurement results are not included in custom feature events. The privacy page explains these separate activities.
Sources & further reading
Our explanations distinguish the browser measurement from the underlying connection. These primary references provide context; they do not endorse this site.
The implemented request sequence, calculations and payload budgets.
Hosting, measurement requests, analytics and optional local history.
Background on browser measurements and Cloudflare endpoints. Our bounded engine is a separate implementation.
Let’s clear things up.
Is this an internet speed test or a website performance test?
It measures the connection from this browser to the measurement endpoint. It does not audit how quickly an arbitrary website renders.
Why does a second run differ?
Traffic, radio conditions, route changes and browser scheduling can affect the result. Compare several runs with matching conditions and methods.
Do I need an account?
No. The public measurement tools and guides work without signing in.