Prepare yourself to be blown
away with the power of OpenAI. You're gonna be amazed with the endless
possibilities that you can use OpenAI. Now, on this video is
gonna be a nontechy guide. That means that anyone can build this
system just by following in this video. And no, it's not hard to do it. So, in this video, I'm going to cover how
much OpenAI costs, how to build it and how to implement it to use it
on something that's actually useful. So, let's get started with this tutorial.
Welcome to my tutorial. My name is George and this is SaaS Master. Let's get started with the most
important part of all. How much is OpenAI is going to cost us? Because, well, we want to know. And to give you a heads up,
it's going to cost us peanuts. It's super cheap to use this. Now, let me show you the pricing table. They use four AI engine models that they have available, which is Ada,
Babbage, Curry and DaVinci. Now, the most powerful one is DaVinci. And for that case, it's the most expensive
one, but it's still super cheap. It's .02 cent per 1000 tokens.
Now, you're going to be asking yourself,
what the hell is 1000 tokens? Well, there's a quick sentence that explains this to us,
which is super easy to understand. So this sentence that I've just
highlighted would cost 35 tokens, which would be the equivalent of
$0.0007, which is super cheap. But if you want to have an estimate of what 1000 tokens would be, well,
it would be a roughly 750 words. So that's practically a short blog. All right, just .02 cents. So, just to give you an idea,
so it's super cheap to use this. Now, they have other AI engines which is
Curie, Babbage and Ada, which are much cheaper and are
recommended for different use cases.
If you want to save, you canva use these. Or if you want to do other types of AI engine writing systems, then these
might be a better fit than DaVinci. And I'll leave a link to the pricing table and I'll leave a link
to this one, the models. And this will explain to you what would be the best use case for the type
of AI engine that you want to use. For example, the Binge is good for complex intent, cause and effect,
summarization for audience, etc. And Curie would be for language
translation, complex classifications, etc. You check this out,
I'll leave the link there. Now, how do we get started with OpenAI? Well, once you are here,
you're going to go into the main page, which I'll leave the link
in the description.
You're going to head on over to API
and then we are going to log in. And the first thing that I want to show
you before we actually build something is how can I cap how much I
want to spend a month? Because we don't want to go crazy or we want to be aware of how much we're
spending because we don't want to go overboard, I don't know, $20,
$100 a month, et cetera.
Head on over to the top page here. And we're going to go into manage account
and we're going to go into billing. So inside of billing,
we have the usage limits. Here's where you're going to set it. So for example, hard limit would be $20. In my case, I don't want to use more than $20 a month because I don't
actually use this a lot. And soft limit $10. So do set that up when
you create your account.
And just to give you an overview of my
usage right here, let me go into usage. This is what I spent. So for example, on the 12 September,
I spent thirty cents, seventeen cents, fourteen cents, five cents,
two cents, et cetera, $0.07. So it's super cheap to use this. Now let's get started
to actually start using this. So what I would recommend is jumping into playground, and this is
where we're going to build it. Now I'm going to show
you how to build this. Now be aware that you can build endless possibilities with this and use
your own language if you'd like. I've built my own AI writing
for Spanish with this. And let me show you how. First of all, let's get started
with the settings on the right. So for example, this is where
we're going to select our model. Remember, depending what we want to use
it for, we selected I recommend DaVinci.
So that's selected the temperature. If you hover over these,
it'll tell you what they are for. I would recommend the most important settings are these, for example, maximum
length, which would be the tokens. For example, if it's going to be for writing AI blogs
and you want to write something really long, well, you want
to give it more tokens. So for example, you canva set
it to 10 tokens to stop there. Now, if you want to respond to reviews, to comments which are more shorter, well,
I would suggest around 200, 300 tokens. Just to give you an idea,
the stop sequence. I'm going to add my stop sequence and I'll
show you where we're going to use this.
So my slack sequence is
going to be this one. I need to hit enter.
There we go. And top P, we can set that up there.
Frequency penalty. I'll set it to one, which would make it not give me
the same results over and over again. So I'm setting to give it
a penalty to do new results. And the best of which is also important because AI is going to take
one time to give you results. But what if you said two? Well, or three? It's going to in the back end do three tests and it's going to give you
the best output out of those three. But heads up, if what you are doing is
costing 100 tokens, we'll multiply it by three because
it's going to do it three times. So it's going to take 300
tokens instead of 100. So just be aware of that.
If you're okay with spending more tokens and get the best results,
do set as many amounts as you want here. Okay?
So just for testing, I'm going to leave it one, and those are the most
important settings. Now let's get into the playground and
where the most important part is about. So let's just say that I want to write
a blog with titles, all right? So let's just say I'm going to tell it, write a whole blog
based on the title, all right? I'm telling you that's what I want to do.
And I'm going to say title,
and this is where I'm going to add it. So let's grab some titles. Let me go into, I don't know, AppSumo. I'll grab something right here
and I'll grab this title. All right, I'm going to grab that there. Let me go into the playground. That's another one. Okay, here's my playground. And I'm going to add the title here. Okay, now I'm going to submit it. So let's go ahead and submit this.
And it's going to start writing.
And it's writing. Write a whole blog based on the title. So it should write a blog
based on this title. So it should be similar to what we have. So here we go. This is what it wrote. Now be aware that we can tweak
this to get better results. Now, if I find that something is wrong,
I can go ahead and edit this. So I don't know, add to it or edit or whatever you want
to do with it and fix it if you need to. But if it's okay, we'll go into the next round of this to make
it even more powerful. So this is where we're going
to use our stop sequence. Remember this one that we added here? Well, that's where we're going
to tell it to stop, all right? And again, we're going to give it another example and we're going
to teach it what we want.
So again, we're going to add title. In this case, we're not going to tell it to write a blog because it
already knows from the beginning. And let's go ahead and grab another,
I don't know, something right here. Blogger. Let's grab this and we'll grab this title. For example, let's go into the playground. Let's add the title and let's
go ahead and submit it again. Okay, and now it's writing
that short block for this. And again, I can add a stop
sequence and do it again. Why would I do this multiple times? Because I want to make it smarter. Okay, in this case,
I could give it different type of results because I want to avoid
getting the same results. Also, what do I mean by that. Let's just say that the first test that I
made was about a WordPress plugin. Well, I don't want to tell this OpenAI
system that I'm building that all the coming posts are going
to be about WordPress. And that's why I would do three to five
of these tests to make sure that are fine and edit what I need
to tweak to make it good.
Now, this is one of the ones
that you can build. Let me go ahead and save this. I'll say short slack on title. Okay, let's go ahead and save it. Let me show you another
one that I've built. So these are the ones that I built. I built another one for hashtags. So, for example, I wanted to give me
three hashtags when I add the post to it. So in this case, this is the post and this is what I'm getting three
hashtags out of it. Let me show you an example post. And again, I'll grab this. So I want it to give me
the three hashtags. Boom.
Just give me three hashtags. Based on that, it could be a long post,
a short post, whatever, and it's going to give me three hashtags
because that's what I taught it to do.
All right, let me show
you another example. Another example that I built is
a thank you reply for YouTube. So, like I told it,
reply to YouTube comment. And if someone says, thanks, man,
well, I'm going to respond to him. If someone says, well, thank you so much,
I'm really glad, enjoy your video, et cetera, and it's going
to give it a reply. So here's a reply. Now, I've implemented this, and if you write a thank you or something
that has to do with thank you on my YouTube channel,
you'll get a response based on AI.
Now, you're going to be asking yourself, well, why would I use this
to reply to a comment? Well, because it's going to be more
advanced, it's going to be more unique than just writing the same thank you
like repeated over and over again. In this case, I'm getting different
type of replies for each one of these.