How To Use GPT-4o (GPT4o Tutorial) Complete Guide With Tips and Tricks

With the release of GPT4O, this tutorial is going to be a comprehensive guide on every new capability that you can use this software for, including some of the new features that were added today. Later on in the video, I'll show you guys some of the things that are coming and the applications that you can do exactly when those applications do arrive.

But without further ado, let's take a.

Look at some of the new capabilities. First off, when you're greeted with the ChatGPT screen, one of the first things you're going to know is that GPT4 is right there. There are three different tiers of models. GPT 3.5, which is the fastest model, GPT4, the advanced model, and GPT4O, the newest and most advanced model. Now, it's important to note that there is a distinction between these models because some people are confused with as to why GPT4O is free, yet it's available to paying subscribers too. The difference between the models, or should I say the difference between the offerings, is that if you're currently paying, paying for GPT4 plus, you have higher rate limits for GPT4O, which essentially just means that you're going to be able to talk to this model a lot longer. If you're using GPT4O in the free section, some of your conversations may be cut short, as this is still expensive for OpenAI, so they can't allow you to use it unlimitedly. But it is something that you can use quite a decent amount. Another way that you can actually access GPT4O is in the playground right now. You can see I'm currently in the playground and, and if you want to access this screen, all you need to do is you just need to go to Playground, then go to the chat screen. Then what you need to do is of course you can see right here you can select your model. You can either select GPT Turbo, GPT 3.5 or GPT 4.0. Now, if you want to do a direct comparison between these models, you can click the Compare button and then you can put GPT4 Turbo here, and then you can put GPT4O here. This is something that you can do and obviously you're going to need a paid account for this, but this is of course just something you can do. Now in addition, it does say system instructions right here. And if you do want to have this, if you already have a preset like I do right here, I can click OK and my loaded preset, I can compare them and essentially a Preset is just basically a way that the AI is going to talk to me. So, for example, if I wanted GPT4 and GPT4 Turbo and I just wanted to remove my preset, I could simply just say, you are an AI assistant, and then I could simply ask it to give me recommendations for a holiday. And you can see here I'm going to click Run. And one of the things that you can see is that on the left hand side, GPT4O is really, really faster than GPT4 Turbo. So one of the key things that you need to know about GPT4 is that it is essentially 50% faster than GPT4 Turbo, which is a remarkable improvement in terms of the usability if you're going to be using GPT4. Okay, try and make sure that you use this model because it's a lot faster than GPT4 Turbo. Now we can go back to the main screen here, because GPT4O actually does have some increased capabilities. One of the things that they didn't talk about, and I'm not sure if you're going to use this, is of course, the coding capabilities on the benchmarks Here with the GPT4O, you can see that there is a huge jump from GPT4 Turbo to GPT4O in terms of the coding capabilities. In order to use the coding capabilities, all you need to do is literally ask it to code something. So I'm going to ask it to code a Python script. You can see right here of the prompts are to use a Python script. And then we can see how quickly it's able to code this. So you can see right here it's telling me how to actually install this, which is what you want to do on your main terminal. You can see it's going to give me some more stuff and then it's going to continue coding. And this is very, very quickly. So if you're Someone that uses GPT4 Turbo to code on a day to day basis, this is something where you know you can use it to do certain things. What I do is, because I analyze charts in my free time, is that you can use it to create very, very simple trading bots. And you can see right here it's able to completely code this from scratch based on your preset variables or whatever kind of things that you do want. You can see right here how you can integrate different things and it shows you how to execute certain trades. And this is something that is very, very effective. You can use this. Now, of course, here you can See that it does say problems, but initially, if you do have any problems, you can just literally copy them, you can paste them back into ChatGPT and then it will give you the updated code and then you're going to be able to use that again. Now, another thing that they actually did speak about was the Mac app. So if you're currently on Mac os, you are able to currently download the latest version of ChatGPT. I will leave a link to download that, because they've left the link to download that in a pretty strange place. But what you're able to do is, because you're able to do this, and I'll leave a link to this short tutorial as well.

You're to able.

You're literally able to, for example, be in a browser, go ahead and take a picture of an image like this person is doing, then copying their images and then bringing up the button to message ChatGPT. Then you could literally say, please code this in Python. And then in a couple of minutes, whilst it's running the code and you just simply wait for it to do so, you're able to get a working code running in Python. You can see right here, this user is able to do this and it's pretty, pretty incredible. So the coding capabilities are something that I've truly never seen before because it's more effective, more faster, and it's pretty native in your app. Another feature that they didn't really talk about, but it was something that I saw mentioned a few times on Twitter, and I did try this on myself, is that you can actually use ChatGPT to generate 3D models. So you can see right here, I'm going to type this in. And you can see, I've said make an STL file of a table with four legs and random attributes. Then you can see right here it is doing its thing, it's currently generating. And all I have to do is simply wait and. And it might take around 20 seconds or so. And then you can see right here, I can literally download this file. And now that I've downloaded this file, I'm going to try and open it to show you guys what it looks like. And this is the file that I literally just downloaded. So this is something that Chat GPT can do. This is Microsoft's pretty basic viewer. But you can see right here that this is a 3D model that it just generated completely from scratch. I mean, I think that this is pretty cool that you can do this. You can do this with other objects as well. I've seen several people do this, and there are really no limitations on what you can do with this software because you could ask it to do other objects. Now, I do want to be honest with you guys, if you do ask this to do other objects, you're not really able to get them as good as possible because it does struggle with certain things that are really detailed. But because we have something that is incredibly simple in terms of the polygons. There's not too many polygons in this. You're able to do this in a very, very quick way because it's just literally five simple objects composing of simple rectangles and just flat square, if you want to call it that, that make up a very simple object. Another secret thing that they actually didn't speak about that you can now do with GPT4O, and I guess you could say ChatGPT, is that you can actually switch between models in the conversation. So let's say, for example, I'm having a chat and for whatever reason I get locked out of the chat. Previously, what would happen is you'd have to go start a new chat with GPT 3.5, but you can alternate between. So with this, it's pretty useful because no longer are you going to be able to get locked out of your chats, because sometimes what would happen is you'd hit your rate limit and then it would say, you know, we can't continue the chat because you spoke too much with this smart model. But what you can really do right here is if you're asking a basic question, you can then, you know, switch it to GPT 3.5 if you want to mid chat. Or you can switch it to another different model and you can see it keeps the initial response here and then you can see it goes to the second one. So if you want to be able to basically see the differences between the models, you can switch between them mid conversation to see how that works. I think that that is something that is really, really cool. Now, what's also cool is that there are also advanced image generation capabilities with this model. They aren't the newer, newer capabilities, but they are upgraded previously from Dall E3, I guess a Dall E 3.5, you could say. OpenAI didn't explicitly state this, but as someone who uses this quite often, I noticed that there is a stark difference in terms of what you're able to do. For example, it's more accurate. So I said, here is an image of the McLaren 650s. This is what I requested. And then I Basically said, can you rotate this image? And it manages to keep a lot more character consistency than before. However, GPT4O, when that image version is actually released, you're going to get a lot more character consistency, a lot more better than you have now. Whilst right now you do have a little bit of character consistency. You can see here, I was playing around with this character and sometimes things do get a little bit lost in translation. In the future you will have a lot more character consistency, but that just isn't there just yet. But what you do have are actually advanced image editing capabilities and I'm going to show you guys how you can do that right now.

So let's say let's make an image of a cartoon cat. So we're going to go ahead and put that in and then we're going to wait for Dall E3 to give us the image. So you can see right here we've got this lovely image of this cat. Now what you want to do is click the image. Then in the top right you can see that there is a select tool. This is where things are going to get interesting. All you want to do now is you want to use this to determine how big your brush is. So essentially what you have at the top left is you have a brush size slider. So you can make this bigger or you can make this smaller. So let's say, for example, once again I wanted to add some sunglasses. I would then adjust my brush size to the size of the eyes. I would then go across the top area like this. Now that I've got this area selected, you can see that it is now highlighted blue. Once this area is highlighted blue, this.

Means that I am about to change things that only exist in this region.

We come to the bottom right with CH GPT. Then we state, can you. Then we state, can you put sunglasses in there?

And you can see that.

Can you put sunglasses in there? We just give it to ChatGPT, we press enter and then now all we have to do is wait for it to update the image. Now this may take a moment, so.

Give it 20 or 30 seconds.

Now one of the things you might need to do with ChatGPT is if you are covering a bigger area, providing ChatGPT with a lot more context is really useful. You can see here that when I did this the first time I it didn't actually work several times. That was because my brush size wasn't that big enough. So, for example, if you're trying to do an area like this and you want to Give it an additional feature. If you do something that that small, ChatGPT might not recognize your selection. If you want something to be pretty accurate, what you're going to want to.

Do is use a bigger brush. If you don't get it right the.

First time, then cover the more of the area so it has context of the Image and then ChatGPT can understand what to put inside the image. Because this system, I'm guessing that it maybe just edits this area right here. So what I'm going to do now is I'm going to try something and essentially what you can also do is you can essentially edit the text. So along here I'm going to put Write the text Mr. Cat along the top right hand corner. So now you can see, after doing that, we can see that it manages to add Mr. Cat in the image. Now, something that you want to understand, like I said before, don't make small things like this and expect it to understand exactly what to add. If you're trying to add something into the image, try to recover as much.

Of the image as possible so that.

ChatGPT is able to change it.

Now, one of the strangest use cases that you also do have is you do have the ability to analyze emotions going on in any said image. And this is because GPT4GPT4O has really, really good vision capability. So this is something that you can use for a huge different amount of use cases. But it's important to note that it's not as good as the updated version yet because certain features haven't been updated. I'm going to show you guys a table that someone made. So to explain to you some of the features that are going to be coming and just a little bit of the confusion, because I know a lot of people are quite confused. Another thing that was recently showed that you can do with GPT4O is really good data analysis. So I don't actually have any current data on hand. So What I've asked ChatGPT to do is I've asked it to actually generate the synthetic data. So I literally just said, can you generate a spreadsheet with synthetic data that I can save as a CSV file? It says, sure, I can generate a synthetic data set for you. Let's create a data set with the following columns. Name, age, yada, yada yada. It generates a data set that I can then download. Now I did check the data over on in Excel and it does look pretty cool, but I'm going to say now analyze the CSV and give me charts and reports. So then I'm going to simply put this back in and then I'm going to head and I'm going to be able to get some nice data outputs from this, which is basically data visualization. So I've given it the file and let's see what it's able to do. It says, here's some reports that we can can do. And you can see here we actually got some very, very interesting pieces of data analysis. This is really, really cool. And it was also showing that recently this is what a lot of people do get paid to do, but it's something now that you can do for either your company, either for your work. And I mean, you can see right here it says country by distribution, age distribution. You can see salary distribution. Also average salary by country. We can see the age versus salary. This is some advanced data analysis that you can do within a couple of seconds and that allow you to identify trends. So for example, you could say what trends. Trends are present. So you could say that. And then of course you could see that you're able to get more different pieces of data by continually asking it different questions. So this is something that is really, really comprehensive and I hope you all find this tutorial very, very useful. So Sam Altman basically says here that the new voice model hasn't shipped yet. If you're currently using this on Android or iOS, the only model that has shipped is essentially the text model of GPT4O. If you're currently speaking with the model, as many people do, that version is still the old version. But when the new version is here, you're going to be able to do a lot more with that. The new version of voice mode with GPT4O in alpha is coming a couple of weeks after May 13th. API access for GPT4O the text and vision currently starts now. However, the API for audio and video capabilities actually do start in the future. And for all Mac and Desktop plus users, that actually does start in the future. However, if you are on a certain version of Mac, you can download it now with the link in the description. So with that being said, the new model is a marginal upgrade over GPT4 and hopefully this tutorial shows you a variety of the new things that you can do with this model, including analyzing images, generating code, and some of the other capabilities with the 4O voice model that are really cool that will be coming in the future.