NEW: GPT-5 Insane OpenAI Model! Cheaper Than Claude Opus 4.1 - FREE On Cursor AI

What's going on guys? Josh Wilcox here. Now, there was some huge news today. And if you're living under a rock, I will fill you in. Chat GPT5 is finally here. The long awaited model that we've all been saying, okay, when is this model coming? People have been asking Sam about it. And now it is finally here. We have GPT5 regular version. We have GPT5 thinking. And then there's even GPT5 Pro. We're going to dive into it. Actually take a look at

Some of the tests. How good is this model in when it comes to benchmarks? And of course, benchmarks don't mean everything. So, we're actually going to use it in an actual real applicable scenario. Try it out in cursor, see what it's made of, and look at some of the cool things other people are doing with it so you can get filled in and start using it today because there has never been a time like this where you can build a business,

Build a software, build really anything um so easily. So, welcome to the age of builders. And without ado, let's dive right into it. All right, guys. So, all links I cover in today's video will be linked down below. As always, take a look here at OpenAI's official uh research post about GPT5. We can see that they say it's their smartest, fastest, most useful model yet with built-in thinging that puts expert level intelligence in everyone's hands. So essentially GPT5 is a unified

System with a smart efficient model that answers most questions and a deeper reasoning model which is GPT5 thinking for harder problems. And then it also has a real-time router that quickly decides which to use based on conversation type, complexity, tool needs and your explicit intent. So GPT5 is their strongest coding model. As you can see here, they just coded a rolling ball miniame. They coded some pixel art right here. Okay, so we can see paint. You can do a typing game,

A drum simulator, lowfi visualizer. And the cool thing is this was all these games right here was just one prompt. So, you know, imagine back in the day to create something as simple as one of these typing games. You know, it took a little bit of time and you know, you could actually use something like this as a service to teach people how to type as a software, right? Different levels and whatnot. Now, you can do things. so simple with these

Tools. All right, it's their top model for creative expression and writing and it's their best model yet for health related questions, empowering users to get informed about and advocate for their health. So, I mean, I don't know about you guys, but I use chat GPT as well as Claude and really any AI mono perplexity, whatever. Even though Perplexity is just a rapper, but um I use that for so many things for my health. I've created gym plans for myself, workout plans,

Meal plans. Um, if someone's maybe like a dog or like a family member is sick or has an illness or if I have an illness, you know, I'll ask AI. So, this is I mean it's good. We can see the trajectory where AI is uh going just with everything and health in what in particular is something that's really going to disrupt it already is disrupting this space, but it's going to continue um being hopefully a really good help for us. So

We can see here in some academic and human evals that it sets a new state-of-the-art across math. So 94.6% on a IME 2025 without tools and then real world coding 74.9% on the SWE verified and then 88% on Ader Polyot. Multimodal understanding is a 84.2% on MMU and health at a 46.2. 2% on healthbench hard. So, I'm not going to spend so much time going through all these specific evals. You can see that there is a lot here. We got the

Coding benchmarks, the humanity last exam benchmarks, all these different ones right here. So, um like I said, links down below if you want to check some of these out more in depth and can see the comparison between some of their older models like 03, GPT40 versus chat GPG5. But we can see here that it is outperforming 03 right here. So faster, more efficient thinking. Okay, we can see for scientific reasoning, for software engineering, it's outperforming it for PhD level science questions.

And then we have a much lower hallucination rate, which is really good. Uh, now I'm in platform.openai.com, which is the back end for developers. If you go to the models section here, we can see that there is GPT5, GPT5 mini, and then GPT5 Nano. So, um they do have different models. Now, on the front end side of things, they actually cleaned up the UI. First of all, they have a nice little uh color um UI right here, which is actually kind

Quite nice. Um I do like like they made the UI look a bit nicer to be honest. Also too, at least for mine, you used to have to like click here and upload a file. Um now, you can actually just drag it over like this, which uh is nice. That's something that Claude could do that uh it was just a little thing, but they're making the UI a little bit nicer. And also too, before they would have all these different models,

They'd be have O40 uh 03, 03 Mini, 4.1, 4.5. And it's kind of confusing, especially like obviously, you know, myself as well as probably a lot of you, you guys know each specific model, their pros and cons, but especially for someone that is just, you know, using it on the day-to-day, they don't know which specific model to use. So, this is kind of good. combines all the pros and cons of these different models into one. And it's very simple. We have

GPT5 and then GPT5 thinking and then pro. Now, if we go to compare models right here, we can see we got GPT5. Let's do GPT5 Nano and then GPT5 mini. There's also GPT5 chat here. Okay. So, as you can see, we it shows the reasoning, the speed of each model. So Nano is very fast, the types of input output it allows, and then the pricing. So they actually made this uh some upgrades too in the back end here. I like this

Compare models feature, especially since they have a lot of different models. So we can see the pricing here on a per 1 million token basis. We got the input of GPT5 right here of $1.25 cash input of 0 uh 313 and then output of $10. Nano is a lot cheaper 5 cents 1 cent and 40 and then mini is 253 and then $2. Okay. All of them have a much larger contact window. So this is actually a huge upgrade. 400,000 window

And then a max output uh tokens of 128 for all three models. We can see the knowledge cut off here. So for GPT5 it was just about a year ago. So September 30th 2024 we can see all the different endpoints. They all have pretty much all of them. Well GPT5 does. These two are missing a couple for distillation and predicted outputs. Now the one thing is if we look at the price of GPT5 comparative to something like Claude right here we

Have Claude Opus 4.1 which if you don't know just released maybe a few days ago and we can see this is their most uh expensive model as well as their top tier model. We have the input of $15 per mill, a output of $75. And even the Claude Sonet 4 right here of $3 per mill input, and 15 per output is still way more expensive than GPT5. So, we can see here $10 per output versus $75 per output and 1.5 or

1.25 per input versus $15 per input. Then, of course, you have cash input, too. But, uh, GPT5 is definitely a lot cheaper. The one thing GPT5 still doesn't beat is Gro 4 on the ARC AGI leaderboard. Now, if you want more updates on GPT5, check out Sam Omen's Twitter, which I'll leave linked down below. He's reposting pretty much different people in the tech industry, basically saying different things about GPT5. Like, we can see Theo here is saying, "I've been using GPT5

For a bit now. The model broke me. It is so good. I didn't know what the prices was. I assumed it would be 03 Pro price because it is that smart." Nope. Truly insane. He literally said he will go into depth for this model. There's a lot more stuff that we could cover. You know, all these benchmarks realistically, they don't mean so much, right? We've seen all the anytime a new model launches, oh, it has this benchmark, this benchmark, this benchmark,

They do mean something. But, uh, let's actually try and build some stuff using cursor. Good thing about cursor right now is for the launch week at GPT5, since if you're on cursor, you get it for a week for free. If I go open up cursor, make sure you're updated to the latest version of cursor. You can see GPT5 and cursor CLI are now available. So try GPT5 for free. It's now available. And they also mentioned their cursor CLI. So bring a

To any IDE or remote machine. So I'm not going to go too deep into the CLI right now. But this is actually very cool because as we've seen with the rise of quad code, Gemini CLI, really every company is doing a CLI version, which you know I do like. There's ADER. There's all these different tools. We're actually just going to go ahead and try GPT5. All right. So, I'm going to create a project here called GPT5 CRM. And we are going

To see what this tool is actually made of. All right. If you don't know how to check for updates, by the way, you just click this uh settings icon, then click check for updates. If there's an update, make sure you update. And then you'll see GPT5 right here. We have GPT5 normal and then GPT5 fast. So, this is the same model, but it's using the fast priority processing, which is typically two times the price. And this one is also offered with

Free credits for paying users during the week of launch. Both of these are medium reasoning. You could also toggle on max mode as well. All right, so to start out, I'm going to use this simple prompt. So, please create me an X.js CRM with all the latest packages, etc. to work all together. I want a nice clean modern UI, light mode, dark mode, and a few key features such as contact management, calendar, task manager, no backend at the moment, just front

End. So, I'm going to use this prompt. We can iterate and improve upon it if need be. Let's go ahead and send this and see what we get. Okay, so now I am just simply going to run things. You can see here that it is going to use nex.js, JS Zustend date FNS React big calendar zod react hook form and sauner for the toast. Okay, so I'm going to run mpm add and then these package libraries right here. I'm going to

Allow this. All right, so packages are now installed. All right, now it's going to run shad cn. So I'm going to allow that. Okay, so now they are getting added right here to the components.json file. All right, so now we have our shad cn in here. Now it's going to scan the uh layout and page.tcxs and then um build everything out. All right. So you can see here it builds out all these different files and now we are just going to

Allow it to rundev port 3000 and let's see what we get. All right. So we have an issue here. I'm simply just going to copy this and put it back in the cursor. All right. Boom. And here we have our CRM. So GBD5 CRM. We can see light mode, dark mode. So, that's good that it's working. You can see contacts, tasks, calendar. So, pretty good right out of box here. Okay. Now, when I click on this right here, it's actually not

Taking me to the page. So, I just let um cursor know that it's not doing that. So, it's working on a fix right now. Okay. So, now it is working. If I go to contacts, we can see the contact page right here to fill out. We can add a contact. We can go to task here. We can see our task right here. add a task, a description, due date, and we have our dates right here, which is nice. We can do

Priority, add task. So, we actually do should be able to add a task like um make video description. Boom. Okay. And now we have our task. We can see we could delete it, add it, etc. We can go to calendar here and we can see the calendar for the month. And this is August 2025. You can see that it did get everything right with the first being on Friday and all the different days. We can see the coloring right here is

A little bit off. We could fix that. I'm not going to bother right now. Now, I'm actually just saying make everything a bit more advanced and cooler with more impressive features and cool stuff and let's see what we get. All right, so cursor aka GPT5 did its thing here and we apparently got some improvements. Let's check it out. Okay, we're getting a build error. Let's copy this and paste it into cursor. Okay, so now it should be fixed. Let's go ahead

And take a look. Okay, so it added a couple new features like I think we can do command K. Okay, so nice command K. And now we can actually kind of search like a command here. Um we can navigate like dashboard, contacts, task, calendar. We're getting a error here. Um I'm not going to bother fixing this right now. It's just a simple error. We hover over contacts, tasks, and calendar. We could see a graph right here, I believe, of kind of

How many tasks and contacts and calendars we have. Right now, we have none. So, it shows zero. If I click on accent up here, we can actually switch the accent to uh red, purple right here. As you can see, it's changing colors. So, now it's purple, now it's blue. We go to contacts here. We could actually uh browse um files to add contacts. We could export as a CSV. I'm not going to do that right now. We go to task here.

We can now see to-dos, in progress, and done. Let's add a test task right here. Now, we can see the test task is in to-do right here. We click next status, and then it's going to go to in progress. Next status. Now, it's in done. So, now that we are back in our dashboard and we have one task, we can see that there is now one task opposed from the others that have zero. a pretty basic CRM, but just wanted to

Give you a sense of how it is out of box with a couple different prompts. Few things I'll say is it did a pretty good job. A lot of the LLMs I use tend to have trouble with uh getting shaden set up with the XJS and there was a couple little things here and there, but it did a pretty damn good job. All in all guys, this model just dropped today and I still want to use it a lot more before

I give a final conclusion as to what I think about this model. But as of now, my first impressions are it is pretty impressive and I am very excited to use it more. Now, I will also be using it a lot more in the front-end user interface right here um within chat GBT, testing it out for different things such as creative writing, different things like that. In today's video, we mainly covered about the coding side using cursor because I know a

Lot of you guys are into that. But definitely let me know what your guys' thoughts are in the comments down below about GPT5. Does it live up to the hype or is it just overhyped and not as great as people think? And if you've been using it, let me know what some of the use cases you've been using it are for. Is it just for coding that you've been using it for? Have you been using it for writing? How do you

Think it compares to Claude? Claude Opus 4.1 thoughts on it 4. And of course, let me know down below if you feel the AGI or not. Other than that guys, that's pretty much it for this video. If you're new to the channel and you got some value here, I appreciate a like, comment, and subscribe to stay up to date with the future videos. I've been super busy on the business side, guys, but I plan to make a lot more videos coming

Soon. I got a lot of cool stuff ready for you guys, a lot of value in the pipeline, so I'm excited to share that with you guys. If you need help growing your business or implementing AI into your business or running an AI agency, go to executivestride.com/apply. Book a call down below with myself or my team and we can see if it's a fit or not. Other than that guys, I will see you in the next video. Check out some of

The links down below. Keep hustling, keep grinding, and of course guys, accelerate your stride. Take care.