Astro is the framework I reach for first, and this site runs on it. It is also the wrong choice for a lot of the marketing teams I talk to. This guide answers one question: when should you pick something other than Astro in 2026, and which of the alternatives are healthy enough to bet on?

The Short Answer

Astro calls itself “the web framework for building content-driven websites like blogs, marketing, and e-commerce.” That is accurate, and it is also the boundary.

Pick something else when one of three things is true:

  • Your site is really a product. Logged-in dashboards, account areas, lots of state living in the browser. Astro can stretch that far, but frameworks built for applications start there.
  • Your marketers need to publish without a developer. On our own site, every article is a Markdown file in a Git repository, checked against a schema when the site builds. That’s wonderful for a developer and a non-starter for a content team that publishes every day.
  • Your team already ships in another framework. A Vue team should look at Nuxt first, a Svelte team at SvelteKit. The best framework is often the one your developers already know.

If none of those applies, Astro is a strong default. If one does, read on.

I’m not neutral here: we build on Astro, our own site runs on Cloudflare, and migrating and consolidating WordPress sites is part of the work I sell.

Who Owns Your Framework Now

In the last two years, “who owns this?” got harder to answer for many options on this list.

  • On January 16, 2026, The Astro Technology Company joined Cloudflare. Its post says “Astro stays open-source and MIT-licensed” and “continues to support a wide set of deployment targets, not just Cloudflare.”
  • On June 4, 2026, VoidZero, the company behind Vite, joined Cloudflare too. Vite is the build tool underneath Astro, SvelteKit, React Router and SolidStart. VoidZero says Vite “will remain open-source and MIT-licensed.”
  • On October 9, 2026, the Deno team joined Cloudflare. The Deno runtime gets one more year of bug fixes and security updates, and then, in Deno’s words, “we will end our development of the Deno runtime.” Deno Deploy, its hosting service, shuts down after six months.

Cloudflare isn’t alone:

So “independent” is no longer a useful filter. Nearly everything has an owner, and every owner makes its promises in good faith.

Deno is the reminder of what a promise is worth over time. On May 20, 2025, Deno published a post saying “We’re not winding down.” On October 9, 2026, the runtime got an end date. I don’t think anyone was lying. Plans change when companies change hands.

The answer isn’t to avoid owned tools. It’s to choose ones you can leave:

  • Plain files. Static output is a folder of HTML, CSS and images that any host can serve.
  • Standard formats. Content in Markdown, or in a database you can export in full.
  • URLs you control. Your addresses and redirects live in your repository, not a vendor’s settings panel.

That is the test I’d apply to everything below: if the owner changed course tomorrow, what would you have to rebuild? How Astro grew, and what the deal means for it, is a story of its own. Here, portability is what matters.

Speed Won’t Decide This for You

The HTTP Archive tracks real-world speed across millions of sites using Google’s Core Web Vitals, its scores for loading, responsiveness and visual stability. In its technology report for September 2026, the share of mobile origins that passed was:

  • Hugo: 78.2%
  • Webflow: 71.3%
  • Astro: 71.1%
  • Framer: 70.8%
  • WordPress: 48.3%
  • Next.js: 35.7%

Before anyone screenshots that: this reflects site type, not the framework. Astro, Hugo, Webflow and Framer mostly carry content and marketing pages. My read is that Next.js runs far more applications, and applications are harder to make fast. Put a heavy app on Astro and I’d expect it to be slow too.

Developer sentiment tells a similar story. In the State of JS 2025 survey, 94% of developers who had used Astro would use it again, against 55% for Next.js, down from 68% a year earlier. But 59% of respondents had used Next.js, against 27% for Astro. It’s a self-selected survey of about 13,000 developers, so read it as mood, not market share.

If speed is your only reason for choosing, almost any content-first tool will get you there, hosted ones included. Choose on who edits the site, how interactive it needs to be, and where it runs.

When Your Site Is Really an App

Next.js has the biggest ecosystem, and for a React product with marketing pages attached, it is the obvious candidate. The cost is portability and patch work.

Next.js made its Adapter API, the official way to run it on other hosts, stable in version 16.2 on March 18, 2026. As of October 2026, its own deployment docs list only Vercel and Bun as verified adapters. Other platforms’ integrations “are not verified… compatibility may vary.”

The security record matters most if you host it yourself:

None of this makes Next.js a bad choice. It means someone on your team owns patch day, and off Vercel, more of it.

React Router is where the original Remix framework went. Remix v2, in the team’s words, “had become a thin wrapper around React Router,” so its features moved in. Version 8 shipped on June 17, 2026, and the team called it a “boring” release and moved to one major version a year. Boring is a compliment for something you’ll maintain for years. Don’t confuse it with Remix 3, which reached npm on October 1, 2026 as a separate framework built without React.

TanStack Start is the type-safe newcomer. Its team declared 1.0 stable on October 13, 2026: “finally stable… ready for production.” It is also days old as a stable release, so I’d save it for teams who enjoy being early.

Nuxt and SvelteKit are the answer when your developers write Vue or Svelte.

  • Nuxt. Nuxt 4 shipped on July 15, 2025, and 4.6.2 followed on October 16, 2026. Nuxt 3 reached end of life on July 31, 2026, so if you’re on it, the upgrade is due.
  • SvelteKit. SvelteKit 3 shipped on October 1, 2026. In the same survey, 88% of SvelteKit users would use it again, second only to Astro.

Both are healthy; use what your team knows.

When You Want Fewer Moving Parts

If developers run your content site and you want even less machinery than Astro, two older tools deserve a look.

Hugo had the highest pass rate of the six in that HTTP Archive data, with the same site-type caveat. It ships a new version every few weeks; v0.168.0 arrived on October 13, 2026. Its lead developer is funded by sponsors, not a platform company. The trade-off is Go’s template language, and Hugo’s own docs admit that “the most common templating errors made by new users relate to context.”

Eleventy calls itself “a simpler static site generator” with zero client-side JavaScript by default. Version 3.1.7 shipped on October 15, 2026. The open question is the rebrand: the next major version will be Build Awesome, alongside a paid Pro product. Font Awesome says the paid tier “will not be required.” I’d take them at their word and still keep my templates portable.

Both are honest alternatives to Astro for a site of pages and articles. Where Astro wins, for us, is TypeScript components and schema-checked content in one toolchain.

When Marketers Need to Publish Without a Developer

For most marketing teams, this is the honest answer: the real alternative to Astro is often not another framework at all.

WordPress is, in my experience, still where non-developers publish with the least friction. Its release pace is steady:

It is open source and runs on the host of your choice, which puts it near the top of the portability ladder. The caveat is governance. On September 24, 2026, a federal judge declined to dismiss WP Engine’s antitrust claims against Automattic, according to Search Engine Journal. I made the longer case, with five tests for any platform, in Why WordPress Is Uniquely Positioned to Thrive in the AI Era.

There is also a middle path: Astro in front, with a headless CMS behind it for editors. A headless CMS stores and serves your content while a separate front end draws the pages. That trade-off has a guide of its own.

Webflow suits designer-led marketing sites. As of October 19, 2026, its Premium plan costs $25 a month billed yearly and holds 20,000 CMS items. Read the exit terms, though. The same page says of code export: “Dynamic content (CMS items/pages) can’t be exported.” Your page designs can leave. Your articles don’t leave with them.

Framer is quick for polished, design-heavy landing pages. As of October 2026, its Pro plan is $30 a month billed yearly, with 2,500 CMS items. On leaving, Framer’s help center contradicts itself:

Both were updated on September 15, 2026. Until that’s settled, I’d plan as though there’s no clean exit.

From most portable to least: WordPress, then Webflow, then Framer.

Check the Pulse Before You Commit

A framework’s homepage always looks healthy. The release history tells you more. Three I’d steer a new project away from today:

  • Fresh, the framework that grew up at Deno:
    • In a June 3, 2026 post, maintainer Luca Casonato wrote that he and Marvin Hagemeister had left their jobs at Deno and moved Fresh into its own GitHub organization.
    • The last release, 2.3.3, shipped on April 28, 2026.
    • Fresh 2 is built on the Deno runtime, which now has an end date. The maintainers plan a Vite-based version that runs beyond Deno, but it hasn’t shipped.
  • Gatsby:
    • The last major version, 5.0, shipped on November 8, 2022. The last release of any kind was 5.16.1, on February 10, 2026.
    • A maintainer describes “limited development”: security fixes, limited dependency updates, small bug fixes.
    • Gatsby’s own homepage invites visitors to “Migrate to Netlify Today.”
    • Only 24% of developers who had used it would use it again. It isn’t dead, but if you run a Gatsby site, plan the move.
  • VitePress:
    • The last stable release, 1.6.4, shipped on August 5, 2025.
    • Version 2 has been in alpha since January 22, 2025, with the latest alpha on September 4, 2026.
    • It works, and Vue teams like it for documentation. Just don’t assume a stable v2 is close.

Treat three more as early:

The Decision Matrix

These are the options a marketing team will realistically weigh. Health is as of October 19, 2026.

Option Best for How it renders Lock-in Health in 2026 Pick it if
Astro Content and marketing sites Static by default; server optional; interactive “islands” only where needed Low: static output is plain files 7.0 on June 22; 7.4.1 on Oct 19; company joined Cloudflare in January Developers own a content site
Next.js React products with marketing pages attached Static, server, incremental, streaming Medium: runs anywhere on Node.js or Docker; only Vercel and Bun adapters verified 16.4.2 on Oct 15; maintained by Vercel; frequent security releases You’re building a React product and want the biggest ecosystem
React Router React apps that can move hosts Server, pre-rendered or single-page Low: templates for several hosts v8 on June 17; yearly majors; developed by Shopify A React team wants an app framework without host ties
Nuxt Vue apps and sites Server, static, or mixed per page Low: deploys to many providers 4.6.2 on Oct 16; Nuxt 3 end of life Your team writes Vue
SvelteKit Lean apps and sites Server, static or single-page per route Low: official adapters for major hosts 3.0 on Oct 1; backed by Vercel Your team writes Svelte
Hugo Large static sites Static Minimal v0.168.0 on Oct 13; sponsor-funded You want speed and no JavaScript framework
Eleventy Simple static sites Static, no client JavaScript by default Minimal 3.1.7 on Oct 15; becoming Build Awesome You want the simplest JavaScript-based generator
WordPress Marketer-run sites with plugins Server-rendered PHP Low: open source, any host 7.1 on Aug 19; 7.2 due Dec 8; governance lawsuit ongoing Non-developers publish daily
Webflow Designer-led marketing sites Hosted Medium to high: CMS content can’t be exported New plans in May 2026; Webflow Cloud runs on Cloudflare Designers own the site and accept platform hosting
Framer Fast landing pages Hosted High: export unclear Pro plan includes 2,500 CMS items A small team needs polish fast and accepts lock-in

The long tail:

  • TanStack Start, SolidStart and Qwik suit specific app teams.
  • Fresh and Gatsby I’d avoid for new work.
  • For documentation, match the tool to your team: Starlight (Astro), Docusaurus (React) or VitePress (Vue).

What We Run, and Why

Our own site took the long way to Astro:

  • This repository began in January 2026 as a React single-page app, with the blog still on WordPress.
  • In March it moved from Vercel to Cloudflare Pages.
  • On August 15 we rebuilt it as a static Astro site and moved the blog into the same repository.

Today a small Cloudflare Worker answers every URL, and nothing falls through to WordPress. A React app was the wrong shape for a site that is mostly words. Static HTML was the right one.

We use Astro as an HTML compiler, not an app framework, and the site loads no UI framework in the browser. Our house build standard says it plainly: “Default to Astro with static output… Do not add a client framework by default.” Canon articles, this one included, are Markdown files with a strict schema, so a missing title or a malformed date fails the build. The feeds, the sitemap, the llms.txt files and a Markdown copy of every article are generated at build time from the same source.

Astro doesn’t stop you shipping weight. It makes you choose it. The particle field across our pages uses an engine of about 300 KB gzipped, and it loads only after the visitor’s first scroll, tap, key press or pointer movement.

On portability, we hold ourselves to the test above. The build output is a folder of files. A redirects file in the repository maps older addresses to their current ones. Only the small Worker is specific to Cloudflare.

The company behind our framework now belongs to our host. I’m comfortable with that because leaving would mean moving a folder, not rebuilding a site.

Two cautions from our own repository:

  • Pin your version. Astro shipped two major versions this year, 6.0 on March 10 and 7.0 on June 22, and 7.4.1 landed on October 19. We pin an exact version and upgrade on purpose.
  • We are not a marketing team. I edit Markdown in Git; your editors shouldn’t have to. For a client whose team publishes daily, I’d recommend WordPress, or Astro with a headless CMS, before what we run.

The full build is documented in our site walkthrough.

Before You Choose

Run the decision through six questions:

  1. Who publishes, and how often? If it’s non-developers every week, start with WordPress or a hosted builder.
  2. How much of the site needs a login or app-like behavior? If it’s most of it, start with an app framework.
  3. What does your team already know? React, Vue or Svelte decides more than any benchmark.
  4. Can you leave? Check that pages, content and URLs export in standard formats before you sign anything.
  5. Is it alive? Look at the date of the last stable release and who pays the maintainers, not the homepage.
  6. What would you rebuild if the owner changed course tomorrow? The shorter that list, the better the choice.

If you’re weighing a move off WordPress, or onto it, our migration guide walks through the questions we ask first.