PNG or JPG for thumbnails
This question gets answered badly all the time, usually as "PNG is better quality". That is true in a narrow technical sense and misleading in practice. The right choice depends on what your thumbnail is made of and how you are going to upload it, and for a lot of designs the two formats are visually identical while one file is five times larger.
What the two formats actually do
JPG throws information away to make the file smaller, and it throws it away in a way tuned for photographs. It is very good at gradual tonal transitions and comparatively bad at hard edges, where it leaves a faint dirty halo.
PNG throws nothing away. Every pixel comes back exactly as you drew it. It compresses very efficiently when large areas share the same colour, and very inefficiently on photographic noise, where files can balloon.
The rule that follows from that
If your thumbnail is dominated by heavy type, flat colour blocks, hard-edged shapes and cut-out graphics, use PNG. Those are exactly the things JPG handles worst, and exactly the things PNG compresses best. You get cleaner letter edges and often a smaller file than you expected.
If your thumbnail is dominated by a photograph — a face filling the frame, a landscape, a product shot with soft lighting — use JPG at high quality. There is no visible penalty, and the file will be a fraction of the size.
Where the difference in file size actually comes from
It helps to know why that rule works, because then you can predict what your own file will do instead of exporting it twice to find out.
PNG compression is lossless and works by finding repetition. A run of identical pixels costs almost nothing to store, and a row that looks much like the row above it costs very little either. A flat colour block covering half your canvas is close to free. Photographic texture is the opposite case: every pixel differs slightly from its neighbour, there is no repetition to exploit, and the file ends up holding something close to all of it. That is why the same canvas can give you a PNG of a few hundred kilobytes for a poster-style design and one many times that for a photograph.
JPG works in the other direction. It converts the image into frequencies and discards the high-frequency information the eye is least sensitive to, with the trade-off tuned for photographs. Smooth gradients are low frequency and compress down to very little. A hard black edge against yellow is about as high frequency as an image gets, so JPG either spends a lot of its budget holding that edge or leaves visible ringing beside it. Photographs come out small and clean; graphics come out larger than you expect and dirty at the edges.
So the two formats have no fixed size relationship at all. For a full-frame photograph the JPG is a fraction of the PNG. For heavy type on flat colour the PNG is often both the cleaner file and the smaller one, and if the design uses few enough colours, exporting PNG with a reduced palette shrinks it further with no visible change. Put a photograph behind that same type and the arithmetic flips straight back.
Most thumbnails are a mix, so what then?
The typical thumbnail is a photograph with huge outlined text on top. That is a genuine mix, and the deciding factor is usually practical rather than aesthetic: how you upload.
From desktop the size ceiling is high, so PNG is comfortable and gives you the cleanest text edges. From the mobile app you have 2 MB, and a PNG of a photographic thumbnail at high resolution will often exceed it, so a high-quality JPG is the sensible choice.
Quality settings matter more than the format
A JPG at a high quality setting and a PNG of the same photographic image are, to the eye, the same picture. A JPG at a low quality setting is visibly worse than either. Most of the "JPG ruins thumbnails" reputation comes from people exporting at aggressive compression to hit a file size target.
If you are choosing JPG, choose high quality and accept the larger file. Reducing quality to save a few hundred kilobytes is a bad trade when YouTube is going to re-encode the result anyway.
The podcast exception
For podcast cover art, the format choice has a hard constraint behind it. Apple's specification says cover art must not contain transparency or an alpha channel, and PNG files routinely carry an alpha channel even when they look fully opaque.
JPG cannot carry one. That makes JPG the safer default for cover art specifically — not because it looks better, but because it makes a whole class of validation failure impossible.
What about WebP and AVIF?
They are better formats by almost every technical measure, and they are not useful here. YouTube's documented accepted formats for custom thumbnails are JPG and PNG. Podcast directories work from the same short list.
Use modern formats for images on your own website, where you control what the browser receives. For platform uploads, send what the platform documents.
What they are, and why they are still not the answer here
It is worth knowing what these formats actually are, if only to understand why they keep being recommended to you. WebP comes from Google and offers both a lossy mode, comparable in intent to JPG, and a lossless mode comparable to PNG, with support for transparency in either. AVIF is newer and takes its compression from the AV1 video codec; at a given visual quality it generally produces smaller files than either of the older formats, and it handles wider colour and higher bit depth as well.
Both are genuinely better tools, and both are still the wrong answer to this question, for two separate reasons. The first is the accepted-formats list. What you may upload is what the platform documents, and no amount of technical superiority makes an unsupported file upload successfully.
The second reason is the more interesting one. Even if you could send an AVIF, you would gain nothing by it, because the format that reaches the viewer is not the format you sent. Your thumbnail is re-encoded into a set of derivatives, and the platform serves whatever suits the surface and the browser asking for it, in modern formats where they are supported. The conversion you were thinking of doing has already been done for you, further down the pipeline, from whatever source you handed over. Your job is to supply the cleanest possible source, not to guess at the delivery format.
Where these formats do repay the effort is on a site you control, serving AVIF or WebP to browsers that accept them with a JPG fallback for the rest. That is a real reduction in page weight and it is worth doing. It simply has no bearing on the file you hand to YouTube or to a podcast host.
Quick answers
Does PNG give me a higher-quality thumbnail on YouTube?
Only in the sense that you hand over a lossless source. YouTube re-encodes it regardless, so the gain shows up mainly in the crispness of hard edges like text.
Is a bigger file better?
Not by itself. A large file from a small blurry source is still blurry. Size only helps when it is preserving detail that genuinely exists.
Which should I use for a design that is only text on a colour?
PNG, without hesitation. It will be sharper and probably smaller than the JPG equivalent.
Why is my PNG smaller than my JPG?
Because the design is flat colour and hard edges. PNG compresses repetition, which is what that kind of image is made of, while JPG spends its budget trying to hold edges it was never tuned for. Put a photograph behind the same text and the two sizes swap round.