How to Use GPT-4V for Web Design

We launched support for GPT-4V in Create a couple weeks ago after OpenAI's Developer Day. We think it will open up new ways to build sites and applications faster than ever before.

If you're short on time, my co-founder, Marcus, posted an epic (early demo) video on Twitter (I mean "X"). What struck us instantly the moment we got it working is that this will allow everyone to build.

Yes, writing code still takes too much time, and at Create, we're working on ways to make it take much less time. But beyond development, often the most time consuming part of the creation process is the design process: deciding what something should look like or how it should work.

That now goes away.

You can quickly turn anything that already exists into ready made inspiration for your next project. Just snap a photo, paste into Create, and voila you have the code for it. On Mac, it's as easy as pressing cmd-shift-4 to copy something to clipboard, then cmd-v in Create's builder.

gpt-4-vision for landing pages

You can iterate from there with further comments, or better yet, further images.

We're finding it works great for creating components you can use in development, turning Figma mocks into ready to deploy code, designing emails, spinning up landing pages, and combining two things into something new.

The possibilities are truly endless.