AxioStats
Explore data
Back to the blog

Journal

The Most Popular JavaScript Frameworks, Ranked by npm Downloads

npm download counts are the closest thing to a popularity vote the JavaScript ecosystem has. Here is how to rank frameworks fairly, what the numbers hide, and the current state of the top five.

Published
Updated
By
AxioStats
Reading time
4 min read

“Which JavaScript framework should I use?” is one of the most asked questions in software development, and one of the worst answered — mostly because the answers are opinion masquerading as fact. The closest thing to a neutral popularity vote the ecosystem has is npm download counts: every framework’s usage is, at some level, a stream of npm install requests, and that stream is public.

This article explains how to rank frameworks by npm downloads without fooling yourself, and shows where the top five stand right now. The live weekly series for five leading frameworks lives in our Framework Weekly Downloads dataset.

Why npm downloads are the least-bad popularity signal

Community surveys are self-selected. GitHub stars measure interest, not usage. Job postings lag reality by years. npm downloads, on the other hand:

  • Require no self-identification. Every install counts, including CI pipelines and npx scaffolding.
  • Are timestamped. You can see trends, not just levels.
  • Are public. Anyone can pull the same numbers (we do, weekly, via the npm downloads API).

That makes download volume the closest proxy for “how many projects actually pull this into their build.”

What download counts hide

If you rank frameworks naively by total downloads, you’ll get a list that says more about packaging than adoption. Three distortions matter:

  1. Dependency trees. A framework that ships 40 satellite packages (helpers, adapters, plugins) multiplies its download count by the number of packages every project installs. A minimal framework scores zero.
  2. Tooling downloads. create-* scaffolding CLIs generate hundreds of thousands of one-time installs that have nothing to do with ongoing usage.
  3. CDN and bundler bypass. Frameworks increasingly served via CDN or bundled into other tools (Astro, Remix, Next generate client islands that may not pull the framework package into your install) undercount usage.

The honest fix: compare first-party installs only — count the core runtime package each framework asks you to install (react, vue, svelte, @angular/core, solid-js), not its satellite ecosystem.

How we rank them fairly

Our dataset uses exactly that principle:

  • One package per framework: the core runtime you actually add to a project.
  • Weekly sums per ISO week, so comparisons across calendar boundaries are apples to apples.
  • The five tracked frameworks: React, Vue, Svelte, Angular and Solid.

The result is a clean adoption signal. As of late August 2026, React leads the field by a wide margin, followed by Vue; Svelte and Angular trade places in the mid-field, with Solid growing the fastest from a small base. The exact numbers update weekly on the dataset page.

Reading the trend instead of the level

A ranking is a photograph; the trend is the movie. Three patterns to look for in the charts:

  • Steep but stable decline — the slow fade of a mature framework no longer winning new projects but still running large legacy bases. High absolute numbers, negative momentum.
  • Sustained compound growth — the tell of a framework winning new projects (the pattern Solid has shown for several years on a small base).
  • Seasonality — January and back-to-school weeks spike; August dips. Don’t read weekly noise as a trend (that is why our series is weekly, not daily).

Not all popularity is equal

Popularity is a filter, not a verdict. The right way to use these numbers:

  1. For hiring — count is a decent proxy for labor pool size in a stack (React’s dominance here is real).
  2. For maintenance risk — popularity correlates with upsream maintenance, security response and ecosystem depth.
  3. For a new greenfield project — popularity is far less important than fit: bundle size, data model, SSR story, and the mental model your team already has. Svelte and Solid are smaller than React by design — they ship less, and that is a feature, not a bug.

A framework with fewer installs but a smaller runtime and a faster build can beat a giant on the metrics that actually matter for your product: Core Web Vitals, time-to-first-byte and developer iteration speed. The matrix framework — what-the-data-is-about on one axis, how-it-was-produced on the other — is exactly how you would organize this comparison, and download trends are the “hard data” engine’s contribution to it.

Explore the live data

Related on the blog:

FAQ

Does React really get more downloads than Vue? Yes — by a large multiple, depending on how you count. React’s ecosystem of satellite packages inflates raw totals further, which is why comparing core runtime installs is the fairer method.

Why is Angular considered less popular than its downloads suggest? Angular’s CLI and tooling generate many installs for a comparatively smaller number of active projects, and much of its long-tail ecosystem ships under the @angular/* scope that we exclude from first-party counts.

Are download numbers a good reason to pick a framework? They are a good reason to consider a framework. Pair the trend with fit for your actual product and team — popularity predicts hiring and ecosystem health, not whether the framework is right for your workload.