Poking around Lynx.js
React-like components, real CSS, dual-threaded rendering. A few evenings in the docs as someone who ships Expo for a living.
My feed is mostly React Native, Next.js, and the occasional "framework X killed Y" thread. Lynx.js barely shows up there.
ByteDance open-sourced it in 2025: TikTok-scale cross-platform UI, React-inspired APIs, real CSS, dual-threaded engine, Rust toolchain (Rspeedy). On paper it sounds like someone rebuilt the interesting half of React Native with lessons from a billion-user app.
I don't do hype posts. I do "I spent a few evenings in the docs and here's what happened."
First impression
The pitch is familiar if you know web dev: write once, render on Android, iOS, Web, HarmonyOS. ReactLynx for components, CSS you don't have to unlearn, PrimJS instead of shipping all of V8.
Getting started is npm create rspeedy@latest. Lynx Explorer on your phone for preview. The tutorials (waterfall gallery, product carousel) are short and actually runnable. That's more than I can say for a lot of launch-week frameworks.
What felt different from Expo
Performance story is threading-first: UI work separated from JS so lists and gestures don't fight your business logic. Coming from RN, where you already think about the JS thread vs UI thread, the model clicks.
CSS is first-class, not a subset you memorize. Animations and layout behave more like the web. For teams with strong front-end chops, that's a real draw.
The honest caveat: Lynx is often embedded (LynxViews inside a native host), not always a greenfield "create app from zero" flow like Expo. If you expect npx create-expo-app and you're in the store next week, adjust expectations.
Why it's not everywhere yet
Ecosystem size. React Native has a decade of libraries, hiring pool, and Stack Overflow answers. Lynx is growing (open-source roadmap through 2026, lynx-ui components, desktop via Clay), but you're early if you bet a client project on it today.
Community discourse is quieter. That can mean less noise. It also means fewer battle-tested answers when you hit an edge case at 11pm.
I'm not switching client work to Lynx. I'm keeping it in the "explore when curious" bucket, same as I do with new bundlers and edge runtimes.
Who might actually care
You're deep in cross-platform UI and tired of relearning layout systems.
You maintain a native app and want web-skilled devs to ship screens without rewriting the shell.
You like reading engine source and want an alternative to the RN architecture debates.
You need proven hiring momentum tomorrow. Stick with Expo or native.
What I'll do next
Run through one more tutorial, maybe embed a small Lynx surface in a throwaway Android shell, and see how painful the bridge feels compared to RN modules I already know.
No prediction about "the future of mobile." Just a note from someone who explores tools before the hype cycle tells him to.
If you're experimenting with Lynx too, get in touch. I like comparing notes with people who read docs instead of threads.