OpenAI’s decision to move from Next.js to Remix was driven by a combination of factors related to performance, scalability, and developer experience.
Next.js is a popular framework for building React applications, but as OpenAI’s projects grew in complexity and scale, they found that Next.js was not meeting their needs in terms of performance and maintainability. Remix, on the other hand, offered a set of features and capabilities that better aligned with OpenAI’s requirements.
One of the key reasons OpenAI decided to switch to Remix was its focus on performance optimization. Remix has a strong emphasis on server-rendering and code splitting, which can significantly improve the load times of web applications. This was a critical consideration for OpenAI, as their applications often relied on large amounts of data and complex calculations that could slow down page rendering.
Additionally, Remix’s data fetching capabilities provided OpenAI with more flexibility and control over how data is fetched and managed in their applications. Remix has built-in support for server-side data fetching, allowing developers to pre-fetch data before rendering a page, which can improve the overall user experience by reducing wait times.
Another factor that influenced OpenAI’s decision to move to Remix was its improved developer experience. Remix provides a more intuitive API and a simpler architecture compared to Next.js, making it easier for developers to onboard and work with the framework. Remix also offers tools for code splitting, caching, and routing that can streamline the development process and make it easier to maintain large codebases.
Furthermore, Remix’s hybrid approach to client-side and server-side rendering was appealing to OpenAI because it allowed them to leverage the benefits of both approaches. With Remix, OpenAI could achieve a balance between fast initial page loads and dynamic client-side interactions, giving their users a responsive and seamless experience.
Overall, the decision to move from Next.js to Remix was driven by a combination of factors related to performance, scalability, and developer experience. By making this switch, OpenAI was able to improve the performance and maintainability of their web applications, while also enhancing the development experience for their team.
thx box
Next js is also really good at being a spa and client rendered javascript on the client. Its not that hard to set up.
Goodbye goodbyes JS
Imagine being in the middle of a nextjs project with openAI, and something new comes along. This is exhausting
The push to render everything on the server is mindless.
Research shows that well over 80% of web devs work on custom enterprise operational software. This use case will almost always be better served, in terms of user-experience, by a dead simple client-rendered, client-routed SPA. And it will be dramatically cheaper for you to operate.
Because Next is just horrible, really clear 😂
I moved from all that React shit back to PHP.
This is just a confirmation to let you know that next js is good, chat gpt has become a very big thing and maintainance would definitely become a very crucial stuff. Unless you have another chatgpt you don't have to worry about this.
Bhag sa-le!!! 🎉🎉
Client rendered but needs Remix? Interesting
Haven't they heard of something called "Nuxt"? 😊
yeh i dont get turbopack either. Vite has been the way to go for some time now.
Nextjs is Remix in disguise
Great video! I have two Next.js app router applications running in production. The app router is quite good but as it renders everything on the server first and then hydrates the client is annoying. This leads to a lot of hydration errors resulting in app crashes even if you mark the component with "use client" directive. Remix on the other hand does it pretty well by exposing actions and loaders (btw next.js pages router was also good at this). Plus next.js is not pretty good with graphql backends specifically where there is a custom authentication system with token rotation and all. The middleware just suck in next.js app router it is worst and can hit the rate limit pretty quickly.
Good they moved to remix!
can somebody give an analogy in hamburgers term or etmli5 (explain to me like im 5) for the part from 3:25 to 4:08 please?
I don't like Remix because of React Router, Next.js's file-based routing is great. Next.js isn't doing extra for SSR, 'use client' and 'use server' are features of React, not Next.js.
More Remix content please!
Because they do not know that Remix announced it would stop being maintained for a looooong time. If they're aware of that, then they didn't migrate to Remix, they effectively pre-migrated to React Router v7.
'cause fuck Vercel. That's why.
As you mentioned, using NextJS for such a client heavy application is like battling against its core idea of server first. I would say that if a web app is super interactive – consider going client side.
Great video!