JavaScript bundles in the build
No external JavaScript bundles are generated. Inline scripts still handle the theme switch and hero video.
Benchmarks
Measurements, and how to repeat them
A benchmark you cannot reproduce is a decoration. This page publishes what this site's own build produces, how Astro differs structurally from the stacks studios usually replace, and the method to get numbers from a project of your own.
What this site ships
Measured on 2026-10-03 from the production build of this repository, with Astro 7.3.5. Sizes are uncompressed bytes on disk, before the compression a host adds in transit.
Stylesheets loaded by the homepage
8,635 bytes shared by the layout and 5,980 bytes for the homepage: 14,615 bytes in total, before compression.
HTML sent for the homepage
29,004 bytes for the complete HTML document in the October 3 build. Stylesheets, screenshots and video are additional resources; this is not the total page weight.
These figures describe this site, which is a marketing site of a handful of pages. They are not a claim about every Astro project: a site that hydrates a heavy island will send that island's JavaScript, and it should.
Comparisons
Astro next to the alternatives
Each comparison sets out what the two stacks do differently, says where the other one is still the right answer, and ends with the method below.
Astro vs WordPress
How an Astro build and a WordPress installation differ in what they send to the browser, what they need from a server, and what a change looks like before it ships.
Read the comparisonAstro vs Next.js
Where a content site and an application diverge: default output, the component model, interactivity and what each stack needs from a host.
Read the comparisonAstro vs Webflow
What changes when a client site lives in a repository instead of a hosted builder: ownership, output, recurring cost and how a change is reviewed.
Read the comparisonMethod
How to measure it on your own project
The numbers that matter to a client are the ones from their own site, on their own content. Five steps produce a result someone else can check.
- 01
Compare the same page, not two different designs. A homepage with a video and a homepage with a heading are not a measurement.
- 02
Look at transferred bytes and the number of requests in the network panel, with the cache disabled and a cold profile.
- 03
Run the audit three times on the same throttling profile and report the median, not the best run.
- 04
Then check field data, which is what visitors actually experienced: the Core Web Vitals technology report is built from real Chrome users and is public.
- 05
Publish the URL, the date, the device and the throttling profile with every number, so someone else can get the same result.
Run the comparison on a project you are already delivering
Astrology opens the client's Astro project, runs your agent on it, and puts the preview, the diagnostics and the diff in the same window as the change.
Requires Node.js, npm and Git · AI features use your installed agent CLI