How thumbnail images are stored
YouTube generates a fixed set of thumbnail sizes for every video and serves them from predictable URLs built on the video's ID. The standard set runs from a 120 × 90 default up to maxresdefault at 1280 × 720, with intermediate sizes at 320 × 180 and 640 × 480.
The maximum-resolution version only exists if the source video was uploaded at 720p or above, which is why some videos return an error or a grey placeholder for it while the smaller sizes work. Older uploads in particular often top out at 640 × 480. If the highest size fails, the next one down is the actual ceiling for that video.
There are also auto-generated preview frames — three stills the system samples from the video itself — which exist separately from whatever custom thumbnail the creator uploaded. Those are the fallback when a channel has not set a custom image.
What you may legitimately do with one
A thumbnail is a copyrighted image, owned by whoever created it, and downloading it does not transfer any rights. This matters because the most common uses are exactly the ones that are least defensible.
Reasonable uses include referencing a video in a review or commentary, where quoting an image alongside criticism is a well-established fair use or fair dealing case; building a link preview to a video you are legitimately embedding; and retrieving your own thumbnail when you have lost the original file, which is a common and entirely unproblematic reason to use a tool like this.
Using someone else's thumbnail as artwork for your own content, in advertising, on merchandise, or as a template that keeps their imagery, is infringement. Thumbnails frequently contain a person's face as well, which brings personality rights into play independently of copyright — a creator's likeness used to promote something unrelated is a separate claim from the copyright one.
Making a thumbnail worth downloading
If you are here to study what works, the constraints are worth knowing. YouTube's specification is 1280 × 720 at a 16:9 ratio, under 2 MB, in JPEG, PNG or WebP. The critical design constraint is that it is displayed at perhaps 210 pixels wide in a search result and smaller still on a phone.
That means faces and large text dominate effective thumbnails for a straightforward reason: they are the only things that survive an 85 percent reduction. Three or four words is the practical ceiling for legible text, and high contrast between subject and background matters more than composition subtlety.
Remember the duration badge sits in the bottom-right corner and covers whatever is beneath it, and that a red progress bar overlays the bottom edge for partly watched videos. Keep both regions clear. Test by shrinking your design to 210 pixels wide and checking it still reads — the same test that applies to social preview cards.