Browser key timing is not hardware latency
True end-to-end input lag starts with a physical action and ends with a visible response. A web page has no independent timestamp for when your finger actuated the switch, so subtracting browser key events cannot measure that delay.
This tool measures dwell time: the interval between a keydown and its matching keyup. It also measures consecutive keydown intervals, sometimes called down-to-down flight time. Both are browser observations of your typing behavior, affected by the input path and timestamp precision.
The median describes the middle sample; the 95th percentile describes a high value in the retained sample. Small samples make percentiles unstable. A key held for a long time can simply reflect the passage, finger movement, or an intentional pause; it does not identify a slow switch.
Timing summaries use the latest 600 events. Incomplete press/release pairs and time across focus loss are excluded from paired measurements. Repeat keydowns are excluded. Keep the page focused and use the same passage and test conditions for comparisons.
Keyboard polling rate is how often the host checks for device reports. Keydown spacing is governed by your presses and the input stack, so its inverse is not a valid USB polling-rate measurement. Hardware probes, controlled actuation and high-speed recordings are more suitable for physical latency studies.
Read how our browser tests work for event handling, measurement limits, and privacy details.