SPEEDTESTER
How fast is your internet?
Check download, upload and responsiveness in one test.
Turn a measurement into a useful next step
Start with a quiet baseline. Change one condition, repeat, and compare responsiveness as well as speed.
Browse diagnostic toolsHow we measureWorked example and comparison worksheet
Connection speed test
Connecting to the test server…
Test server and observed network
Server: Automatic. Requests are routed to an available service endpoint. This is not a discovered server inside your ISP.
Observed public network: Not identified. A VPN or proxy can change this identity.
Network information refreshes on reconnect, tab return and before and after each test. No current network data.
Manual server selection is not connected: additional independently hosted and verified test endpoints are required. No ISP nodes or server locations are fabricated.
Test settings Up to 4096 MiB
Quick checks allow up to 1 GiB. Transfers stop at the time limit or your data allowance, whichever comes first.
Higher data allowances help fast connections collect enough samples. How the test works
Test speed chart
Waiting for first measurement
Network assessment
Evidence-based, not speculative
Activity tests.
Select the activities you want to compare, then run them together.
Build your network scenario
Select what each simulated device is doing. These generate real traffic from this browser; they do not connect other physical devices.
Test configuration
Configured automatically from your chosen activities and simulated device counts.
Select an activity to configure a test.
Select activities to calculate the test duration.
Calculated from combined traffic demand, pressure duration, protocol overhead and a reserve. Includes two capacity checks of up to 6 seconds each. Relayed video media can use additional data.
Capacity sampling allowance
Fast connections can exhaust a small allowance before enough capacity and loaded-response samples are captured. Raise this limit and repeat if that happens. Activity demand, pressure duration and accuracy requirements stay the same. A larger allowance uses more test data and does not guarantee a qualified result.
Assumed concurrent demand, not measured utilisation.
Beyond the speed test.
Check this browser, and learn which network details need a direct connection to your equipment.
Upload and upstream responsiveness
Send up to 16 MiB of test data while measuring round-trip responses before, during and after upload. This tests this device’s path to the test server; it does not inspect individual ISP hops.
Browser-only device responsiveness
Check whether this browser stays responsive over three seconds. It runs on your device without transferring test data. It does not measure internet speed or system CPU usage.
What we can see about your connection
Your network path
What we can and cannot directly observe
Cloud measurements show the end-to-end path. Independent router, mesh and ISP-node measurements require authorised integrations.
Available checks and measurement limits
Internet path
Effective throughput, baseline responsiveness, loaded RTT excess and measured browser main-thread scheduling pressure.
Gateway & router
Separate WAN from LAN performance, queue depth and network-device resource use where authorised.
Mesh & Wi-Fi
Mesh parent links, wireless backhaul, radio retries, link quality and per-satellite diagnostics.
ISP infrastructure
Access-node utilisation and direct provider-side faults require actual operator telemetry.
Mesh Testing.
Walk between locations with the same device. Compare your internet path before and after changing your mesh arrangement.
Locations, node notes and connections are your labels. The browser cannot identify your access point, inspect backhaul or test the router-to-curbside link. Rates measure sequential HTTP application goodput, including request overhead, not physical link capacity. There is no weighted mesh score or certification.
1. Your comparison plan
Test history.
Results are saved in this browser only. Nothing is sent to a reporting account.
How to read these measurements
Clear boundaries produce trustworthy results.
Directly measured
Bytes received from the test server, uploaded bytes acknowledged by the server, browser-observed round-trip times and latency during transfer phases.
Modelled, not impersonated
Workload profiles evaluate aggregate synthetic bandwidth demand and responsiveness targets; they do not contact games, exchanges or AI providers.
Not yet measurable here
Router-to-ISP speed, mesh backhaul, radio statistics and physical ISP access-node congestion cannot be determined from a standard browser alone.
Internet speed, latency and Wi-Fi guides
A speed test is a snapshot of the path between this browser and a measurement server. Compare download and upload capacity with responsiveness under load, then use repeatable checks to investigate a problem.
Keep the evidence behind your comparison
Record your device, location, test settings and background traffic. Repeat the original setup after a change to check whether the difference persists. Keep incomplete results with their reasons; do not substitute zero for an unavailable measurement.
- Measure your original setup (A).
- Change one condition and repeat (B).
- Restore the original setup and repeat (A) before drawing a conclusion.
Download the blank A/B/A comparison worksheet (CSV) to record your runs. The Wi-Fi comparison guide explains what a room-by-room result can establish.
Read the worked engineering example and inspect its redacted measurements (JSON). This is one historical engineering run with its recorded method and limitations, not independent calibration or a typical-speed benchmark.
Frequently asked questions
Why are some speed or latency results unavailable?
A valid metric needs enough verified observations and sustained sampling time. A short window, exhausted data budget or insufficient RTT probes produces an inconclusive result rather than a guessed speed.
Can this test identify my router, mesh backhaul or ISP fault?
The browser measures the end-to-end path. Direct router, radio, mesh and physical ISP observations require authorised integrations that are not connected in this public beta.
Are these real games and streaming services?
No. Workload profiles generate synthetic traffic against controlled endpoints. They do not certify third-party services, and one browser cannot reproduce contention among several physical Wi-Fi devices.
Are my results uploaded to a reporting account?
Results stay in this browser unless you export them. Our hosting provider processes the network requests and temporary synthetic data needed to deliver the site and run tests.