Test · Train · No sign-up
Test your reflexes.
Practice your aim.
Visual and audio reaction tests, plus browser aim and decision practice.
Test your reaction time
Visual · 5 trials · median
Start visual testAim training in Arena
BetaStart with Flick practice. Explore five separate fixed-mode benchmarks.
Open Arena →Compare like-for-like on the same setup. Visual timing is browser-estimated; audio includes uncalibrated output delay. Arena metrics are task-specific, not a reflex-test score.
Practice aim and decisions.
Start with two sets of 24 Flick practice targets, or explore five fixed-mode benchmarks. Flicking, tracking, choosing and stopping are different tasks with different metrics — not one combined reaction-time rank. The engine loads only after play intent in Arena.
Practice and fixed-mode competition have different purposes. Arena complements the reaction-time test — it doesn't replace it.
Targets pop, you snap — aiming throughput per second.
Glue the reticle to a target that won't hold still.
Right answer, right now — the cost of choosing.
Cut the wire on go, hold on the bomb — how well you stop.
Read the center arrow while the flankers scream the opposite — focus under noise.
Aim · tracking · choice · go/no-go · conflict response · ranks provisional
Understand what you measure.
Reaction tests report a browser-observed interval, not isolated hardware latency or a universal ability score. Read the method and compare only compatible sessions.
Painted-frame visual timing
Visual onset is timestamped in a browser frame callback — inside requestAnimationFrame — not a timer alone. The browser-resolution estimate scales with your display: about ±8.3 ms at 60 Hz and ±0.8 ms at 600 Hz. It does not include unobserved input, panel or compositor delay.
Honest about audio
Browser audio output latency is unknowable to the millisecond — and brutal on Bluetooth. We label audio results as relative, warn on wireless, and never pretend an absolute number is exact.
A scoring core you can audit
Stats, percentiles and anti-cheat run in a small Rust core compiled to WASM — the same code on your device and on our server, so the leaderboard re-scores every run from raw trial data.
Find your baseline.
Visual ranks by median. Audio ranks by 10-cue average. Both keep uncertainty visible.