NEW OpenAI GPT 4.1 Update is INSANE (FREE!) 🤯

Today we have the new release from GPT4. 1 in the API. This literally just came out a few hours ago like you can see right here and this is a new series of GPT models featuring major improvements on coding. Now this is the first thing to note about the whole update is that is specifically focused on coding and this is specifically in the API. Now if you actually go to chat GPT from what I can see it's not actually available inside there. So if we go on the drop down you can see inside the list there is no 4. 1 there. There's cheapy 4. 5 still but there is no 4. 1 which is weird but basically this is mostly focused on coding. I want to show you today how you can build anything and also how you can get free access to this model because despite it being an API you can't get free access and I'll come on to that later. They've basically

said, "Today we're launching three new models in the API. GPT 4. 1 Mini and GPT 4. 1 Nano. " So, there are three different models here, and these models outperform GPT40 and GPT40 Mini across the board with major gains encoding and instruction following. It also have larger context windows. From what I've tested, it's ridiculously fast at replying, but we'll come on to that later. And if you're wondering, okay, how does this perform on benchmarks? So you can see for coding GPT4. 1 scores 54. 6% on sw bench verified improving by 21. 4% over GPT40 and 26. 6% 6% over GPT 4. 5 which is interesting right so it's a lot better than cheap 4. 5 for coding as you can see right here in terms of instruction following on scales multi-challenge benchmark a measure of instruction following ability GPT41 4. 1 is superior over GPT4 all right so there's some big improvements here in terms of long context on video MME a benchmark for multimodal long context understanding GPT4. 1 one scored 70 72% on the long no subtitles a 6. 7% improvement over GPT40. So basically this is a huge improvement over GPT40 and apparently according to the benchmarks it's also outperforming GPT 4. 5 trying to figure out okay what does this mean in terms of what like where are we going in terms of the numbers here cuz we're going up and down like a yo-yo GPT 4. 5 then 40 and then 4. 1 etc. But here we are my friends. All right. So the GPT4. 1 family intelligence by latency. You can see the differences here. So you got GPT4. 1 Nano at the bottom and then you have GPT 4. 1 the main model over at the top over here.

Now if you want to get access to this, one of the easiest ways is just go over to open router and you can actually add a model. So you can use GPT 4. 1 and then you can add different models over here. If we have a look at the latest models, we've got three different options. So, we have 4. 1, 4. 1 mini, 4. 1 nano. Now, you can see the differences in prices here. So, 0. 1 per million input tokens, 0. 4 and then $2. All right, so you can see the prices right there. And these support a million token context window, which is absolutely huge versus previous context windows from OpenAI. And here are the prices. This is also available inside root code client open router of course kilo code etc. So if you do want to use it if you like chat GPT inside the chat then just go to open router find the model that you want to use cuz obviously there's three and then you can click on chat over here and decide which one you want to use it and I'll also be testing this against claude 3. 7 as well in a later video. So let's keep going through now. So these

improvements in instruction, following reliability, and long context comprehension also make the 4. 1 models considerably more effective at powering agents. Now 2025 is the year of agents. And I can understand why they're starting to release models just for coding because that's where everything is going. If you look at the way the world is going, everything is revolving around agents, building apps, building tools, etc. And this just probably is more optimized for that, more efficient for that, and more powerful for building out better quality agents with chat cheap. What else we got here? They've

also said that they're going to begin deprecating GPT4. 5 preview in the API. And the reason for that is that GPT4 offers improved or similar performance on many key capabilities, but it's lower costs and it's lower latency. All right, so GPT4 preview is actually going to be turned off in 3 months. Something to bear in mind if you've built stuff already around GT4. And what's interesting here is they've said we'll continue to carry forward the creativity, writing quality, humor, and nuance you told us you appreciate in GT4 into future API models. Now, they've also broken this down with alpha testers. This is really interesting

because Windsurf have actually offered free API usage of chat GPT for. So, if you go over to Windsurf, I'll be showing you how to build stuff later today, but if you go into Windsurf over here, and I'm on the free plan, you'll see from the drop down, you can select all these different options. And the main one you want to focus on is GPT4, and that is free for a limited time to write and chat directly inside Windsurf, which is absolutely awesome. So, if you want to get free access to this, if you want it for free using chat GPT, then you can use Windsurf. I may compare this as well against Gemini 2. 5 Pro as well as Claude because it's gonna be interesting to see. My personal favorite for coding right now is actually I would say it's a tie between Claude 3. 7 Sonet and Chat GPT sorry Gemini 2. 5 Pro and Claw 3. 7 Sonnet. Gemini is typically a lot faster but you hit the limits faster as well. And if you can code with all three for free for now with Windsurf as well and that's absolutely awesome, right? So let's keep going through. Now we've got

the S sw bench verified accuracy cheapy 4. 1 is right at the top there in terms of accuracy and this is a model basically where a model is given a code repository and issue description and must generate a patch to solve the issue right performance is highly dependent on the prompts and tools used blah blah but basically this is outperforming all the previous models on the same benchmark test and apparently this is much better for editing large files the good thing about this is you could use it locally you could build with visual studio code or windsurf because it's edited seem like a large set of files, it should be better and more accurate at doing that. In terms of the ADA Polyglot benchmark accuracy, you can see GPT4. 1 right here. This is actually not outperforming Open. So, just bear that in mind. And 03 Mini as well is still outperforming GPT 4. 1 in terms of polyglot accuracy. Let's keep going through now. Here's an example of chat GPT40 versus GPT 4. 1. So basically this is a prompt to make a web application flashc card right so it's a flashc card web application the user should be able to create flash card search through their existing kind of like an app style and you can see the differences in the UI right here so if you have a look here the UI is just a lot nicer on this side versus this one this is 40 this is 4. 1 and the quality is better right this one is a lot more simplified when you actually look at the responses right there so we got some real world examples so windsurf have said winds Surf is kind of like Visual Studio Code. It's similar as some better features, but we'll come on to that later. Windsurf have said GPT4. 1 scores 60% higher than GP on Windsurf's internal coding benchmarks, which correlates strongly with how often code changes are accepted on the first review. The users noted it was 30% more efficient in tool calling and about 50% less likely to repeat unnecessary edits. Right, so it's basically just a lot better coded, more efficient. Here's another one from Kodo. instruction following supposed to be a lot better. This is a huge page by the way if you look like there's so much to this. There's so many different benchmarks and statistics and that sort of thing which is really interesting. And then you can see here we've got the other models. But yeah, basically this is it. Now if you want to see the reaction, what's the reaction so far? So this is the tweet from Sam Alman. Again, this just came out like literally a few hours ago as you can see. Sam Almuner says GPT4. 1 and mini and nano are now available in the API. These models are great at coding, instruction following, and long context. 1 million tokens. Benchmarks are strong, but we focus on real world utility and developers seem very happy. GPT 4. 1 family is API only. But again, if you want to work around that, if you want to start using it inside the chat, then just go to open router. All right. So all the video notes by the way inside the AI profit boardroom as you can see right here and if you go inside open router you can start using it and you can even connected to web search right here. So if we say okay what happened today 15th April 2020 we're using GPT 4. 1. If we hit enter that should be connected to the web and we should be able to use this as a chat model not just for APIs. All right as you can see it started finding the details here and you can use it inside the chat. is super fast when it replies. It's absolutely awesome for that. Thomas Rotors have also been using it for AI assistance, that sort of thing. MMU accuracy, so here's some more benchmarks. GPT4. 1 is outperforming 40, but OpenAI 01 is still outperforming it, right? So, these models are still more powerful. GPT 4. 5 is on par for MMU accuracy. And then you got mass vista accuracy over here. So, it's tied with pretty much everything. The 4. 1 and 4. 1 mini are performing pretty well right there. And here we go. Now, if you're wondering, okay, what's the pricing? Here's the comparison of each. So, in terms of where is it now? Where's that pricing gone? Here we go. All right. So, model, you got 4. 1, 4. 1 mini, and 4. 1 nano. Right. So, nano is way cheaper versus 4. 1 mini versus 4. 1. So the full model like full power is pretty expensive but if you use a nano version it's cheap as absolute chips right there. And then overall we've given a nice little conclusion which is cheap 4. 1 is a significant step forward in the practical application of AI by following real world developer needs. All right. So let's see what it can do. Let's take this bad boy for a spin. We're going to have a cheeky gander. Now I'll use it for free inside

wind surf cuz I know that's what the peeps want to see at home. We're going to open it up. Here we go. I don't know why open. It gets a lot of hate these days to be honest with you, but we're just going to give this a cheeky. So, I'm going to open up a folder over here and we'll set up a new folder and we'll just try and create like a 3JS runner game. Just see what happens. All right, we'll grab one of the prompts from the AI profit boardroom cuz that is the place to go for prompts, my friends. So, what we can do is if we go down here and we'll take an Optimus Alpha prompt. Let's see what we got. We're going to create a dinosaur game. We'll grab this prompt. We'll go back into Wind Surf inside the right section. So, you want to make sure you use right and then also select GPT4. 1. By the way, use dark mode in Windsurf cuz light mode is absolutely terrible. All right. And then we're going to say make me a captivating endless runner game. Key instructions on the screen. P5Js scene. No HTML. I like pixelated dinosaurs and interesting backgrounds. We're going to give that a cheeky world. Now, we'll hit enter and we'll see what we get back. So, it's going to start generating now and we'll see what we get back in a second. You can also do the same inside open router chat. So, if you go inside open router again, depending on which one you select, I would go for nano if you bit of cheapkate on a budget. Go inside the chat here. Plug in the prompt and then hit enter and you're good to go on that, my friends. So, it's beginning to generate that. It does say the request is taking longer than expected. I can imagine so many people have jumped on to Windsurf and just started rinsing it for credits, but we seem to have got that back. All right, so we've got the P5 JS here. Let's grab that. Then what we're going to do is we'll go over to P5 JS editor and we'll just hit play on that. And there we go. We have the game ready to go. Let's see what we got here. Oh, it's incredible, isn't it? It was pretty fast to be fair. I'm not going to lie. I wouldn't lie to you. Like that is really fast. and we just built a game for free using p5 js and winds surf and you know exactly how to do it. If you want the prompt for that, go inside the AI profit boardroom. It is the place to be and prices are going up this month. So make sure you sign up now before you miss it.

That's basically it. That's how you can use this whole model and it's pretty simple and easy to build. We'll build some more stuff in a sec. But yeah, there'll be more videos coming on like how to build things with chat cheapy and I'll put all the prompts inside the AI profit boardroom. If you actually want a community that's focused on making money, saving time, and growing with AI, growing your business, feel free to get the AI profit boardroom. It's an amazing community. Comes with all my best SPS, tips, templates, workflows, etc. Like you can see right here. And if you want to get my workflows for basically building anything with Chat GPT 4. 1, you can get that inside here. It's got like many different SPs like you can see tons of different SPS including how to use wind surf BDIY client directly in the chat made. com and R code with chat cheapity 4. 1. Also you can ask in the community any questions you have and additionally just jump on the weekly Q& A calls which also get recorded if you can't make them. So make sure you sign up now before you miss out. And if you want a free one-to-one SEO strategy session that shows you how to grow your business with SEO, feel free to get that. We'll show you how to retail websites from zero to 145,000 visits a month and generate hundreds of thousands of dollars in sales on autopilot. And on this free link building acceleration session, you'll get a custom tailored game plan to save you hundreds of hours with SEO and get you more free leads, sales, and profits using this. All right, so you can jump on the call. It's based on what works for all our clients. Jump on the call, ask any questions you have, and you can book that in directly inside the chat