GEARCHECK LABCompare response delay before and during controlled download and upload traffic. This is a bounded HTTP loaded-latency comparison, with a maximum of 64 MiB of test payload per load direction.
Starts only when you choose it. May transfer up to 64 MiB down and 64 MiB up (128 MiB total payload), plus small probe responses. Load phases target about eight seconds; fast links may hit the data cap and produce an inconclusive sample.
No probes sent yet.
Baseline uses successful small HTTP probes before the load. Loaded medians use probes that overlap active transfers, and additional delay is loaded median minus baseline. A negative difference is possible because the path varies over time; it is not negative hardware latency. Each load phase targets about eight seconds and stops at its payload budget. At least five baseline probes and three overlapping probes per direction are needed for a complete comparison.
The endpoint is the GearCheck Lab server, not an automatically chosen nearest server. A fast link may consume the data cap before enough overlapping probes are collected, so the correct result is inconclusive rather than a fabricated grade. Browser connection limits, server speed, VPNs and the network path can also affect delay. This tool does not guarantee line saturation or certify router queue behavior. Stop cancels pending transfers; a hidden tab interrupts the run.
The traffic may have ended too quickly, a transfer may have failed, or too few probes overlapped the load. A fast connection can exhaust the data budget before useful samples are collected.
Diagnostic data is processed locally in this tab and is not intentionally sent to analytics. Network tools send only their test probes and temporary payloads to GearCheck Lab’s speed-test server. Media and device access starts only after your explicit action and any required browser permission. Downloaded results stay on your device.
Read the privacy policy and testing methodology. If something fails, report the tool URL and visible message.