Understand performances concerns with xstyled.

One of the reason xstyled was created is performance. It was originally a drop-in replacement for styled-system which was not performant due to its internal algorithm. Since the beginning xstyled is focused on being as performant as possible.

Impact on bundlesize

xstyled is around ~15kb gzipped including all utilities and all dependencies.

For ~15kb it includes everything to build an entire website or design system.

Impact on performance

xstyled has a very small impact on performance, it is heavily optimized. Every property computation is cached. Every single part of the code is optimized to reduce the footprint as much as possible.

Most of time using xstyled could lead to a performance improvement of your website. This website is entirely build with xstyled, it scores 99 on Lighthouse performance audit.


Does it make sense to create namespaces with less utilities than the x.*?

No, xstyled iterates on component properties, not on namespace utilities.

Does it make sense to use styled.* instead of x.*?

To be honest, we didn't measure it but we didn't have any complaint about it.

15kb is huge! Can I reduce it?

No, it's the cost of the productivity dear. It includes every utilities, for a comparison, Tailwind including everything is 294kb gzipped.


Benchmarks are available in xstyled repository. I ran it on a MacBook Pro (2,3 GHz 8-Core Intel Core i9, 32 GB 2667 MHz DDR4).

xstyled vs styled-system

This benchmark measures the speed of @xstyled/system vs styled-system.

@xstyled/system x 431,828 ops/sec ±1.06% (86 runs sampled) styled-system x 549,224 ops/sec ±0.85% (90 runs sampled) Fastest is styled-system

Why is xstyled slower?

xstyled and styled-system basically does the same thing. They are both heavily optimized to be as fast as possible. The small difference is due to several things:

Even if it is slower, it is still very fast and it should not have any impact in your application. This documentation website has a 99 Lighthouse performance score 🤗.

Styled components evaluation

This benchmark measures the speed of styled components evaluation. It impacts code parsing when your application is loaded.

styled-components x 45,910 ops/sec ±1.49% (86 runs sampled) @xstyled/styled-components x 17,538 ops/sec ±2.10% (87 runs sampled) Fastest is styled-components

As you can see xstyled is slower. xstyled uses RegExp to detects properties in CSS that's why it has a small impact when your component is evaluated. However this impact is negligible compared to the hydration or mounting phase operated by React.

Styled components rendering

styled-components x 1,008,161 ops/sec ±1.24% (86 runs sampled) @xstyled/styled-components x 1,009,807 ops/sec ±1.13% (86 runs sampled) Fastest is @xstyled/styled-components,styled-components

xstyled has no impact on rendering.

Edit this page on GitHub