This is a web performance benchmark of image formats. AVIF came out 51% smaller than JPEG on our test photo at the same 1920 px width. WebP came out 39% smaller. Support is no longer the excuse: 95.35% of users can see AVIF.
We took one 9.8 MB camera photo (4964 × 2782), resized it to 1920 × 1076, and encoded it with the same engine at Auto quality:
| Format | Bytes | vs JPEG |
|---|---|---|
| JPEG q85 | 183.7 KB | +31% |
| JPEG Auto | 140.3 KB | base |
| WebP Auto | 85.5 KB | −39% |
| AVIF Auto | 69.3 KB | −51% |
Yet almost nobody ships it
The 2025 Web Almanac found only 0.7% of LCP images in AVIF. WebP support sits at 96.74%, AVIF at 95.35% (caniuse, August 2026 data). The gap between "supported" and "shipped" is mostly build tooling.
Three things that trip people up
- Quality numbers don't transfer. AVIF q80 is not JPEG q80. On one test image AVIF q80 weighed 159.5 KB, while JPEG q80 weighed 82.1 KB. Tune each format on its own scale.
-
<picture>order matters. The browser takes the first type it supports, not the smallest file. Put AVIF first. - Measure decode on a cheap phone. We measured bytes, not decode time. AVIF decode costs more CPU, and that lands in LCP's render delay.
The setup we ship
<picture>
<source type="image/avif" srcset="hero-800.avif 800w, hero-1920.avif 1920w" sizes="100vw">
<source type="image/webp" srcset="hero-800.webp 800w, hero-1920.webp 1920w" sizes="100vw">
<img src="hero-1920.jpg" width="1920" height="1076" alt="..." fetchpriority="high">
</picture>
Are you serving AVIF in production yet? If not, what's blocking you: encode time, CDN, or the CMS?
Method and the full comparison: https://blinkimg.com/blog/avif-vs-webp-vs-jpeg?utm_source=devto&utm_medium=social&utm_campaign=avif-webp-jpeg

Top comments (0)