WEBPery

Real-world WebP file sizes versus JPG and PNG at 1920×1080. Typical savings, what affects them, and how to benchmark your own images.

WebP File Size Benchmarks: Real-World Numbers

WebP's headline claim — 25–34% smaller than JPG, ~26% smaller than PNG — holds up in practice, but the real numbers depend on image content and quality settings. This guide gives representative file sizes at common web dimensions, explains what moves them, and shows how to benchmark your own images honestly.

For the format itself, see What is WebP?. To produce your own comparisons, convert with JPG to WebP.

How much smaller is WebP than JPG?

WebP is typically 25–34% smaller than JPG at equivalent visual quality. The saving comes from VP8's block prediction, which JPG's older transform-only encoding lacks. On photographic content the relationship is stable, though the exact percentage varies with image complexity.

A representative photographic comparison at quality settings tuned to match visually:

Image (1920×1080)JPG q=85WebP q=80Saving
Landscape photo~420 KB~285 KB~32%
Portrait photo~310 KB~210 KB~32%
Detailed cityscape~560 KB~390 KB~30%

Numbers are indicative; your content will vary. The comparison detail is in WebP vs JPG.

How much smaller is WebP than PNG?

WebP lossless is around 26% smaller than PNG for the same image, and lossy WebP can be far smaller again for photographic content stored as PNG. Graphics with flat colour and hard edges see the lossless saving; photographs wrongly saved as PNG see dramatic lossy savings.

Image (1920×1080)PNGWebPSaving
UI screenshot (lossless)~480 KB~355 KB~26%
Logo with transparency (lossless)~120 KB~88 KB~27%
Photo saved as PNG (lossy WebP)~3.1 MB~240 KB~92%

The last row is why photographs should never ship as PNG. See WebP vs PNG.

What affects WebP file size the most?

Quality setting and image content affect WebP size more than anything else. Lower quality and simpler content produce smaller files; high quality and fine detail produce larger ones. Resolution scales size roughly with pixel count, and the alpha channel adds a modest amount.

The main factors, in order of impact:

  1. Quality factor — the single biggest lever for lossy WebP.
  2. Image content — flat colour compresses far smaller than fine texture.
  3. Resolution — file size scales with total pixels.
  4. Compression mode — lossy versus lossless; see Lossy vs Lossless.

How do you benchmark your own images?

Benchmark by converting a representative sample at matched visual quality, then comparing bytes. Pick images that reflect your real content — a hero photo, a thumbnail, a graphic — rather than a single test image, because savings vary by type.

Convert the sample with JPG to WebP or PNG to WebP, or script a batch comparison with cwebp. Compare WebP at q=80 against JPG at q=85, which are visually comparable.

Where to go from here

WebP's 25–34% saving over JPG is real and consistent on photographic content, and its advantage over PNG for misplaced photographs is enormous. Benchmark your own sample to see the exact numbers for your site.

WebP and Open Source: libwebp Licensing

Is WebP open source and royalty-free? The libwebp BSD licence, the additional patent grant, the WebM repository, and what it means for commercial use.

WebP and PageSpeed: Serve Next-Gen Image Formats

Google PageSpeed flags images that should use next-gen formats. What the warning means, how WebP fixes it, and how to verify the savings.

WebP on iOS and macOS: Safari Support

WebP works on iOS and macOS since Safari 14 and Big Sur (2020). What Safari, Preview, Photos, and Mail support, and older versions.

WebP in Email: Is It Supported in HTML Emails?

WebP is not reliably supported in HTML email. Which clients render it, why most do not, and what formats to use for newsletters instead.