Skip to the check

Frame rate test

Five seconds, every frame counted. Including the bad ones.

Keep this tab in front. Every browser throttles hidden tabs, and a throttled result describes the throttling rather than your machine.

Live frame rate fps

One bar per frame appears here.

If the number is low

Hardware acceleration is off

A standard troubleshooting suggestion that people apply and then forget. With it off, the processor draws every pixel and everything becomes jerky in every site.

Next step: Chrome and Edge: Settings → System. Firefox: Settings → Performance. Turn it back on and restart the browser.

Something runs on every frame

An extension, a tab with video, or a page with an animated advert competes for the same budget as this test.

Next step: Open a private window with extensions disabled and run the test again. A large difference names the culprit.

The device is saving power

Laptops and phones lower the frame rate on battery, and some halve it outright when warm.

Next step: Plug in, let the device cool, run it again. Compare the two numbers rather than judging one.

What this counts, and what it does not

5 seconds
Test length
frames the browser gave this tab
What is counted
median, not average
Headline figure
worst 1% of frames — where stutter lives
Second figure
median 50 fps and worst 1% above 30 fps
Verdict “smooth” at
no — see the limits below
Measures your graphics card

Other ways to measure frame rate

Approach Best for Trade-off
The overlay in a game Gamers, because it measures the actual game engine on the actual hardware Only works inside that game, and each engine counts slightly differently
The frame counter in browser developer tools Developers debugging a specific page Buried in a panel, and the tools themselves cost frames while they run
A moving-picture refresh rate test Comparing 60, 120 and 144 Hz monitors by eye Answers a different question: how the screen behaves, not how many frames the browser produces
This page Anyone whose browser feels sluggish and who wants a number rather than an impression Measures the browser path only, and browsers cap page animation well below what a fast monitor can show

When this page will not help

Where this stops being the right tool — and what to use instead.

Words you will meet on this page

Frame rate
How many pictures per second are drawn, in fps. Above about 50 movement looks continuous; below 25 the eye starts to see individual steps.
Refresh rate
How many times per second the monitor updates, in hertz. A different thing from frame rate: a 144 Hz screen showing 30 fps still looks jerky.
Worst 1%
The slowest one per cent of frames. This is what people call stutter, and it is invisible in an average — which is why it is shown separately here.
Hardware acceleration
Letting the graphics card draw the page instead of the processor. Switched off, everything in every browser becomes jerky, and this is the most common single cause of a low reading here.
Throttling
A browser deliberately slowing a tab that is hidden or a device that is on battery. Not a fault; it is why a test must run in the foreground.
Compositor
The part of the browser that assembles the final picture. It sets the ceiling for page animation, and it is why fast monitors often still read 60 here.
V-Sync
Synchronising drawing with the monitor’s refresh so frames are not torn in half. It also pins the frame rate to a multiple of the refresh rate, which is why readings cluster around 60 and 30.

Frame rate questions people actually ask

Why does my 144 Hz monitor show 60 fps?
Because browsers cap page animation, usually at 60 frames per second, independently of the display. That is a browser design decision, not a fault in your monitor or your graphics card. Games bypass it; web pages do not.
What is a normal result?
On a healthy machine, a median around 60 with the worst frames above 30. A median near 30 usually means hardware acceleration is off or the machine is busy; below 25 something is genuinely wrong.
Why show the worst one per cent?
Because it matches what people complain about. “It stutters” is almost never about the average — it is about the handful of frames that arrive late, and an average of 58 fps can hide frames arriving at 12.
The number dropped after I opened more tabs.
That is expected and is a useful result. Tabs running video, timers or heavy scripts compete for the same resources, and a browser extension doing work on every frame is a common hidden cause.
Does this test my graphics card?
No, and any page claiming to would be misleading you. Browsers limit page animation, so a strong card and a modest one usually score the same here. This measures whether your browsing is smooth.
How do I check whether hardware acceleration is on?
In Chrome and Edge it is in Settings under System; in Firefox under Performance. Turning it off is a standard troubleshooting suggestion that people then forget to undo, and it halves results here.
Does the test send anything?
No. It counts frames the browser hands to this page. There is no request, no permission prompt and nothing stored beyond a one-line verdict in your own browser.
Why five seconds?
Shorter runs catch a burst rather than a pattern, and longer ones simply annoy. Five seconds is enough to produce a few hundred frames, which is what makes the worst-one-per-cent figure meaningful.
It says almost no frames arrived.
The tab was throttled — you switched away, or the browser suspended it. Run it again with the tab in the foreground and do not change windows.
My phone reports lower than my computer.
Usually power management. Phones reduce the frame rate when warm or on low battery, and some enter a low-power mode that halves it outright.
Can this explain why video stutters?
Partly. If the reading is low, video will stutter for the same reason as everything else. If the reading is fine and only video stutters, look at the connection instead — that is a different fault with a different fix.
Does the moving square matter?
Yes. Without something actually being drawn, a browser is entitled to hand out frames lazily, and the measurement would describe an idle page rather than one doing work.