I would strongly recommend Vue3.

I am also an old timer. Hand rolled HTML and JavaScript since '97?

Came up with ASP (both VBScript and JSCript (it's like we're full circle with JS SSR)). Move to ASP.NET when it was in beta. jQuery, Prototype, Knockout.js.

Vue.js SFC is the thing that feels closest to HTML + JavaScript components done right. It's reactivity model is the same as that of JS whereas React's reactivity model is "inverted" [0]

The state model (Pinia or just composable reactive primitives) is simple, Just Works, and never gets in your way.

Lots of folks like JSX because they don't like HTML and CSS. I'm the opposite; I like Vue because I like HTML and CSS and I like JavaScript. Vue is HTML + JavaScript done right with a congruent reactivity model.

[0] https://chrlschn.dev/blog/2025/01/the-inverted-reactivity-mo...

My use of Next.js is due to SaaS products that only support Next.js on their first party support and teams I work with, thus I won't be using Vue, regardless of how great it happens to be.

Privately I seldom use any kind of JavaScript frameworks, other than when doing WebGL/WebGPU side projects, and then it is mostly Babylonjs.

I think Vue 3 is better than Svelte 5's runes, but unfortunately, in my 7 years of using Vue, I've realized that Vue is just a follower of React and may never catch up, as it is only trying to optimize the status quo rather than break the deadlock.

The exact opposite is probably true at a technical level.

Vue 3.6, for example, is incorporating a new signals rebuild that's several times faster and probably the fastest signals implementation out there[0]

Vue Vapor will also release in experimental mode in 3.6 and once it's mainstreamed, will probably also be among the fastest render models.

Both changes do not affect the DX and require no migrations, but will provide faster performance for ever more complex apps.

[0] https://www.youtube.com/watch?v=zvjOT7NHl4Q

Yes, that's exactly my point. Vue 3 just improves things using existing technologies instead of creating truly fresh solutions.

Wait, wait: so your take is that the upcoming alien-signals[0] library integration -- created from scratch to improve Vue's signals performance -- is just "using existing technologies instead of creating truly fresh solutions"?

Basically, if I understand correctly, it's not "shiny" enough -- even though it is going to significantly improve signals performance.

[0] https://github.com/stackblitz/alien-signals

vuejs is 10 years old

And it has changed a lot in that time. The code we wrote in 0.x is very different from the paradigm of 3.x, and while I agree it's largely a change for the better, it's still a chore to keep up with it if that isn't the only thing you do.

The change from V2 to today is much smaller. So small in fact that the paradigm of v2 is now a simple extraction under the hood of v3.

V2 came out in 2016. The number of apps written before v2 is minuscule compared to even newer JS frameworks of today.

I would argue that Vue has been one of the most stable frameworks in JS world and that stability and focus on simplifying the state mental model (aka, it just works) has been the reason for Vue’s continued steady growth across organizations and projects of all sizes.

I'm not really sure if I can agree on stability. For all its warts and flaws, you can largely still write React 18.x the same way you wrote React 0.x. There's many new features and fancy tools in your pocket, but underneath it all, you can take the knowledge you had of 0.x and go straight to writing 18.x. It might not be whatever is the current vogue best practices, but it will work. While React has had many breaking changes over the years, I would argue that all of them are fairly minor, whereas Vue, Angular and now Svelte have introduced big breaking changes. Stability wise I feel like React comes out on top in that regard.

I do agree that Vue is clearly the better paradigm, but unfortunately it's still very much lagging behind React in terms of adoption. I think big breaking changes might have a considerable part in that, it's hard to sell that upwards.

And in 10 years I only took part in a single project using it, back in 2019.

I would put Vue on third place of FE JavaScript frameworks.