JPEG XL vs. AVIF for web use-cases
jakearchibald.com
October 16, 2026
49 min read
57/100
Summary
JPEG XL is now supported by all major browsers, but benchmarks run on a 2024 MacBook Pro with an M4 Pro found AVIF generally more efficient for lossy web images. Across tested photos, the best JPEG XL files were typically 15–25% larger than equivalent AVIF files at default encoder settings and 6–20% larger at maximum effort, although some images—including a weasel photo—favored JPEG XL. The newer jixel JPEG XL encoder generally outperformed libjxl, except on detailed transparency. AVIF produced substantially smaller files for screenshots, illustrations, and images with alpha channels, while also usually decoding two to three times faster at medium quality. AVIF 4:4:4 is recommended as the default over 4:2:0, whose reduced color resolution rarely cut file size and could limit quality. Higher effort settings improved AVIF and jixel results, whereas libjxl’s higher efforts often added large time and memory costs for little benefit. JPEG XL remains stronger for lossless compression and can transcode existing JPEGs without further quality loss, reducing their size by 9–20%. It can also provide progressive previews in Chrome and Firefox, while progressive AVIF currently renders only in Chrome. AVIF supports HDR gain maps in browsers; JPEG XL’s format supports gain maps, but browser decoders do not. For typical web delivery, AVIF 4:4:4 at a practical high-effort setting was the recommended default, with jixel worth testing for unusually large AVIF results.
Key Takeaways
- AVIF usually produced smaller lossy web-image files than JPEG XL, particularly for screenshots, illustrations, and transparent images.
- JPEG XL was the clear winner for lossless images, with lossless AVIF 30–120% larger than JPEG XL in the tested photo cases.
- JPEG XL can convert existing JPEGs without additional loss and reduced tested JPEG file sizes by 9–20%.
- AVIF generally decoded faster, while JPEG XL commonly required two to three times as long at medium quality.
- AVIF supports browser-usable HDR gain maps, while JPEG XL gain maps lack browser decoder support despite format-level support.
What the discussion said
The discussion is almost entirely about web-image compression rather than AI or machine learning, so it offers very little usable AI-specific sentiment. The one AI-adjacent thread uses LLM Arena as a model for how codec claims could be evaluated: several readers favor large-scale blind human comparisons over declaring victory from a handful of benchmark images. Their underlying point is familiar to AI evaluation debates: a metric can reward systems that look strong on a curated test while missing failures obvious to people in real use. Readers also raised the broader problem of optimizing around perceptual scoring systems. One participant argued that an encoder tuned aggressively to a particular similarity metric can produce results that score well yet look worse to human observers, especially on difficult or unusual images. That maps directly onto concerns about benchmark gaming and out-of-distribution robustness in ML, but it was offered as an analogy rather than a claim about an AI model. No commenters debated model capability, training, safety, deployment, or AI access directly. As a result, the thread cannot support a substantive positive or negative assessment of AI/ML sentiment.