There's a new AI coding app called Windsurf that's been really hyped lately, and it's basically been popping up

everywhere. So I just wanted to test it myself and really find out if all of these things are true: Is it really

better than Cursor? So yeah, I found a couple of main differences, and that's what I want to talk about in this video.

So let's jump in. Okay, so the first thing I actually want to cover is the UI. So you can see that both of them actually

look pretty similar, and that's because both of these programs are forks of VS Code, and so they have very similar

features. But a few of the features, of course, are different, especially when it comes to the AI features. So the first

thing that you'll notice, which I actually find quite cool, is when I open up Cascade here, which is basically the

Composer equivalent of Windsurf—it's their AI agent feature. If you open that up, you can see that they have this "write" versus "chat" toggle, and I actually really like that because in Cursor I often have the problem that I want to explore the

code and just ask some questions and just use AI as a sparring partner. And often when I do that in Cursor, it just

starts suggesting code for me. So I think Windsurf actually solves this really well in having these two different modes.

So the "write" mode is like the normal way how you would also use the Composer to write your prompts and then have it

execute the prompts and write the code for you, whereas the "chat" feature is more for you to have this chat with your code

base in the context of it. So here, for example, I'm saying, "Hey, explain what this project does," and it just tells me about

the code base and analyzes it, so gives me some information about it. And importantly, it doesn't suggest any code.

So it's just a good way for me to talk to an AI assistant here. Yeah, I really like this. I think they solved this

really well. All right, the next big difference is how Windsurf actually handles the context. For example, here I

opened up the Cascade WR mode and pasted this initial, big copy-code prompt. And then what it does here is it says,

Okay, I'll help you create this. And then the first thing it says is, 'First, let me check the current project structure.' It

basically does this analysis and reads whatever is inside of your project at this moment, and then it goes really

like step-by-step. So I really like how it's breaking things down. And it says, 'Okay, first let's configure the next...', uh,

do config, and then let's create these components here. And then it analyzes the layout file. So it's really cool how

it goes step-by-step. And this is much more of how you would expect an assistant to work, right? You, as the human,

should just get a high-level overview of what's happening, and the assistant should be updating you on the different

steps that it's taking. Whereas with Cursor, it goes more directly into action. So if you look at how the composer works

here, it's the same prompt, and it says, 'Yeah, no problem. I'll help you do that,' and it just jumps straight into

implementing code, and it doesn't have any of these more agent-like behaviors where it would analyze something, analyze

a page, and then starts actually implementing code in that page. So another really interesting thing about the UI in Windsurf is that the terminal

is, by default, integrated. So Cascade is able to run terminal commands, which, of course, Cursor previously was not able to,

and the default Cursor composer mode is still not able to, but the newly released agent mode is able to do that as well.

And if you compare how Cascade uses the terminal, it does feel more smoothly integrated into the workflow than with

Composer Agent. So let me show you how the terminal is integrated into Cascade. Let's just set up a Next.js and

shadcn skeleton, and let's just run this. Okay, so you can see here it says, "Suggest terminal commands." It suggests

this, and then you can just say, "Do you want to run this command?" Accept or reject, right? And it's waiting on

a response, and you can click accept, and it will run that. So really simple to use. Obviously, there's this added security layer of

that the terminal commands need to be approved by you, which is good. Now, so let's have a look at how this looks in

the Cursor Agent.  Feature: set up a Next.js and shadcn skeleton. Okay, so it's actually similar.

Run command. You can run that, and then what I don't like so much about the UI here is

that you actually... right now it's loading everywhere; it's showing loading everywhere, so I have no idea what to do,

and it's actually waiting for me to make my selections here, but I need to... before I can scroll here, I need to click into

this. Okay, so I need to click into this and then scroll to the right and realize, "Oh, I need to hit Enter," uh, and keep

going here. Okay, the UI here is just slightly off, but yeah, I'm sure the Cursor team is going to improve that soon. So

yeah, overall, I think the Windsurf UI just feels a bit more intuitive and easier to use than Cursor currently. Um,

so definitely a plus point here for Windsurf. Okay, let's go to the second comparison category here, which is going

to be about code quality. Both tools, of course, use the same underlying models; in this case, I'm obviously using CodeSonnet.

The code quality should be similar, but I actually found in my tests that Cursor tends to produce higher-quality results

in my code, for whatever reason. But yeah, that's what my tests have shown. So let's go through an example, and for that I'm

going to use these Copy.ai prompts, and we're going to build out something using these prompts and just see and

compare these two projects with each other and see if the Cursor version looks better or if the Windsurf version

looks better. Okay, so I finished the implementation here in Cursor. What we basically did here was we took these

two Copy.ai prompts and I pasted them into Cursor in the normal Cursor mode, not in the agent mode, and yeah, built out

this front end of this application. If you want to see a more detailed process, you can go look at my Copy.ai

tutorial video, but that's basically what we did, and this is the result that we got. So if we click through the pages, you

can see they all look pretty good. This was also all in one shot, so I—there were no build errors. I didn't have to correct

anything; I just had to go through this process, and then all of the pages were built for me. So Cursor did a really good

job here. So let's see how it is for Windsurf. Okay, now we're here in Windsurf,

and I did the same thing in Windsurf. I pasted in the first Copy.ai prompt,

and it built out the initial part of this implementation, and there was—I think, yeah—there was one little error

which I had to paste in and then fix it. And then I pasted the second prompt, and interestingly, what it did was it

actually went ahead and built out everything in one go. So you can see it created all of the different pages in

just one goal, so truly like an agent that kind of just fulfills the entire task list; so build out all of these

things. But then there were a few problems, so it didn't look great and I had to iterate on it, try to fix it, and

often it didn't really make a difference after I gave it the feedback, so I had to go in loops a few times, and

that really took me a couple of tries here until it improved a few things. But at the end of the day, if we look at the

result, so this is what the Windsurf build looks like. If we click into the pages here, these look empty compared to the

Cursor build. So overall, I would say the Cursor build definitely looks a lot fuller and more complete, and uh, Cursor

did a better job here. Okay, and the final category that I want to make a comparison in is the pricing. So if we

look at the pricing pages here, both obviously have a free version, and it's fine to use that, but you'll see that

both of them don't allow any premium usage of premium models in the free version. So in Cursor, you will get uh

2,000 completions and 50 slow premium requests. So these slow premium requests are use of premium AI models, but they're

slower requests; it'll take you a bit more time. And then in the $20 per month pro version, you actually get 500 fast

premium requests. In Windsurf, you get five premium user prompt credits, so you can only use the premium models five

times, which is really not a lot. And yeah, I don't know about you, but I only use Claude, and I don't use any other models;

I think that gives me the best results, so I only want to use premium models, and which previously used to be $10 per

month, and now they increased it to $15 per month, which is the pro version, which is very similar to the pro version of

Cursor is just a bit cheaper, where, yeah, you get 500 premium model user prompts, so very similar to what Cursor has there, so...

Overall, how does Cursor and Windsurf compare? I would say it's very interesting how they're positioned. I think...

Windsurf is positioned a bit more on the side of, tailored towards beginners, people that are not experienced developers, whereas Cursor is a bit more for...

experienced developers that obviously still want to use AI in their process. Windsurf might be a bit better for...

beginners. I think it's not totally on that side of the spectrum, like GitHub Copilot, that really is tailored very much for...

total beginners, people that have absolutely no idea about code and just want to very easily prompt something and create code. It is a little...

bit simpler to use than Cursor, but if you really want to take the route of creating something more complex and...

creating something that is uh, production-ready, that you can monetize, I would still stick with Cursor for now. It seems to...

be producing better code quality, at least in my tests, so that would be my recommendation for now. And if you would...

like more support, and even if you're a beginner who has never written code before and would like to start building...

something with AI, we have a community here called the Prompt Warriors Community where there are over 1,000...

builders. We're building things together. We have regular calls, we have chats with other builders who are...

helping each other out, so come join us there if you want to learn how to code with AI. All right, bye-bye.

Windsurf vs. Cursor: Which AI Code Editor is Better?

Scripts