About BlinkBench
Know what you measured.
Take a reaction time test online and you get a number and a percentile: you were faster than 63% of people. Two things go unsaid. Which people — measured on what hardware, in what sample, filtered how? And how much of your result was you at all, rather than the delay between the screen changing, your browser noticing, and your keyboard reporting?
That second question has a real answer, and it's bigger than most people expect. A 60Hz display can only change every 16.7 milliseconds, so the stimulus you reacted to may have been waiting up to a frame before you could possibly see it. Add the input path and the browser's own event timing and a meaningful slice of any sub-100ms “reaction time” belongs to the equipment, not the person. Eric built BlinkBench because no test that hands out percentiles was willing to say so.
Eric builds free, single-purpose measurement tools — a network of sites whose shared rule is that a number has to be defensible before it ships. BlinkBench applies that rule to a category that mostly doesn't: browser tests that score people against samples they never describe. He is not a psychologist, and BlinkBench does not diagnose, train, or rank you against a population it can't document. It measures carefully and tells you what the measurement is worth.
Part of the Veraino network of free websites.
How it works
Most tools that show their work are showing you a formula. A test is different: it's an instrument reading, and the instrument is your browser, your input device, and your display — none of which we control. So showing our work here means four things, on every test.
- The raw measurement comes first. The number you actually produced, in the unit it was measured in, before any interpretation.
- The apparatus limits are stated on the page. Your display's refresh rate quantizes when a stimulus can appear at all; the input path and the browser's event timing add more. On a short measurement that is not a rounding error, and we show it rather than absorbing it silently into your score.
- No percentile we can't defend. Where published research gives a reference range for a comparable task, we cite it and put it beside your score as context. Where it doesn't, you get no percentile at all — not one invented, and not one computed from our own unvetted visitor sample.
- No claim that the score is you. A reaction time measured through a browser is a fact about a person and a machine. We won't pretend otherwise, and we don't sell training to improve it.
Our commitments
- Free, no sign-up required. Every tool on this site is free to use with no account, subscription, or paywall.
- Your scores stay on your machine. We don't store results, we don't build a leaderboard out of them, and there is no account to make. Nothing you do here becomes a row in a table we own.
- Honest about limitations. Where the measurement is limited by the equipment or the method, we say so on the page rather than in a footnote. These tests measure; they don't assess, diagnose, or predict anything about you.
- Kept current. Every cited reference range carries a “last reviewed” date and is re-checked against its source on a schedule — an ongoing responsibility, not a one-time build.
Get in touch
Questions, corrections, or suggestions: hello@blinkbench.com. Eric reads every message personally.