Decode Speed Test
Same images. Different decoders. See how quickly your browser turns compressed files into pixels.
Put your browser to the test
5 images · all formatsFive photographs, 2400 pixels on the long side, each tested as the lossless PNG master plus 13 encodes: JPEG, JPEGli, WebP, AVIF at 8/10/12 bits, and seven JPEG XL variants. Encoding settings vary by format; these files are not matched for visual quality.
Preparing the test…
Keep this browser tab visible while testing; switching tabs cancels the test run. Results appear together when the run finishes.
Your results
| Image | Encoder | Decoder | Pixels | File size | Median time | Speed | Status |
|---|
Decode speed by image
Megapixels per second · higher is fasterAll bars use the same scale, sorted fastest first. When both JPEG XL decoders run, the thinner WASM bar sits over the full-height native bar, both starting at zero; the number shows native speed when your browser has it, otherwise WASM. Hover over or focus a bar for both speeds in MP/s.
Native JPEG XL native JPEG XL WASM
What are we measuring?
From bytes to pixels
Files are downloaded before timing begins. Each result gets one untimed warm-up, followed by sequential decodes. We report the median time and calculate speed from the decoded pixel dimensions. Downloads, loading the WASM decoder, and drawing to the screen are excluded.
Native decoding uses a fresh image and object URL for each decode. WASM creates a fresh decoder and copies its output into JavaScript memory as 8-bit RGBA. These timings include the work of the respective browser APIs; they are not isolated codec CPU measurements.
Two ways to decode JPEG XL
If your browser decodes JPEG XL natively, the test uses that. You can also include our self-hosted jxl-rs v0.7.4 WebAssembly decoder for comparison; it is slower and significantly lengthens the run. Without native support, JPEG XL always runs through WebAssembly. Both decoders get exactly the same bytes, and unsupported or failed files stay visible in the table.
Results depend on your device, browser, and encoding settings. Background activity, thermal throttling, and browser scheduling can affect timings. Compare multiple runs; this is a local experiment, not a universal ranking.
Inspired by Jon Sneyers’ browser decode speed test.