Test a YouTube thumbnail at the size people actually see it
A thumbnail is designed at 3840×2160 in a design tool, judged at full screen, and then met by an audience at a fraction of that. The gap between those two viewings is where most thumbnails are lost — not because the image is bad, but because the decision was made at a size nobody browses at.
So this page skips the mockup and renders your image at the sizes a browser actually gives it. Those sizes were measured against YouTube on 13 September 2026 rather than copied from a blog post: 500×281 in desktop search, 420×236 in a channel grid, 410×231 in a narrower search window, 375×210 on a phone, 325×183 in the watch-page sidebar, and 230×129 in that sidebar on a 1366-wide laptop. The smallest of those is the one shown first, on purpose.
Drop a second image and the two render side by side at every size, which is the comparison that is hard to do by eye and easy to do here. The checks alongside them are the ones that can be answered from the file itself: pixel dimensions against YouTube's 640-pixel minimum, aspect ratio against 16:9, and file size against both of YouTube's caps — 2 MB if you upload from the phone app, 50 MB if you upload from a computer.
Nothing leaves your machine. If you need the current thumbnail off a video to compare against, the YouTube Thumbnail Downloader will fetch it, and the Image Resizer will get an odd-sized image to 16:9 before you test it.
What size should a YouTube thumbnail actually be?
YouTube's own guidance asks for 3840×2160, 16:9, JPG or PNG, with a minimum width of 640 pixels. That is what you upload. It is not what anyone sees. Measured in a browser on 13 September 2026, the same thumbnail renders at 500×281 in desktop search, 420×236 in a channel grid, 375×210 on a phone, 325×183 in the watch-page sidebar on a 1920-wide screen and 230×129 in that same sidebar on a 1366-wide laptop. The last one is 3% of the pixels you uploaded, and it is where a large share of impressions happen.
Is the limit 2 MB or 50 MB?
Both, depending on where you upload from. YouTube Help states 2 MB from the mobile app and 50 MB from desktop, for the same video. Almost every thumbnail guide online quotes 2 MB flat, because that used to be the only number — so people compress a good image into a bad one to clear a limit that does not apply to them. This page tells you which of the two your file clears.
Why does the text on my thumbnail vanish?
Because you designed it at 1280 pixels wide or more and a lot of people meet it at 230. Text that is comfortable at full size is about a fifth of that height in the sidebar. The fastest check is the one this page leads with: look at the 230×129 rendering first, and if the words are not legible there, they are not doing any work.
What is covering the bottom-right corner?
The duration badge. Measured on YouTube's mobile site, it is 30.5×18 CSS pixels, inset 12 from the right and 9 from the bottom of a 375-pixel-wide thumbnail. It uses the same 12-pixel type on desktop, so it stays roughly that size while the thumbnail grows — which means it takes more than twice as big a bite out of a 230-pixel sidebar thumbnail as out of a 500-pixel search one. Anything you put in that corner is smallest exactly where it is most likely to be hidden.
Is my image uploaded anywhere?
No. The file is read by your browser from your own disk and drawn into this page. There is no upload request to watch for — open your browser's network tab and you will see the page make none. That also means nothing is stored, nothing expires, and you can use it on an unpublished thumbnail.
Does comparing two thumbnails here count as an A/B test?
No, and it is worth being clear about the difference. This is a preview: it shows you what two images look like at the sizes that matter, side by side, before you commit. A real test needs real viewers, and YouTube has one built in — Test & Compare takes up to three variants, runs in YouTube Studio on a computer, needs advanced features enabled, and picks the winner on watch time rather than clicks. Use this page to decide which two are worth testing, then test them there.