LXBod7UDRqE

Let's learn how to use Google AI Studio (gemini 3)

Watch on YouTube ↗

Transcript

[00:00] in today's video we're going to be checking out everything we can do with Google AI studio if you don't know what that is think of it like open ai's dashboard co-pilot's dashboard this is as a developer our way of accessing AI whether or Google therefore we're going to learn everything we can do within this dashboard when it comes to pricing creating prompts and basically what this means for us when we create software or integrating it within no code Solutions like zapier let's go and get started and jump into today's video I am excited y'all we have a nice little dashboard now that it's dedic ated for Google AI

[00:31] and its API and everything we can do when it comes to leveraging it in software automations this is fun stuff here want make sure to leave a link to this in the description down below I will also leave a link to everything else I reference in this video such as the pricing and a quick little cookbook of different problems we can start using let's go ahead and get going here and sign into Google AI Studio once we're in the dashboard here we have a couple of different options here so anything below that little dash line is purely for educational purposes and just giving you more insight of how to leverage this platform right off the bat if you want

[01:02] to just start leveraging the API key in automation softwares you can simply hit get API key once you're here you can go ahead and click create API key now if you want to see an entire video I dedicated for that topic check out that right there as I go over how do you integrate an API key into zapier this could apply to other automation softwares and on top of that the advantageous fact that it's actually free to use this kind of API key in that context if you don't even know what API key is the purpose of it is that it allows you to access the endpoints we found in the conversational model like Gemini I and they know how to charge you

[01:33] basically this is you I'm going charge you you use me now that we're on the topic of API keys and getting charged money let's figure out pricing coming over to pricing we have a couple different models kind of like what we saw with open Ai and Chad gbt 4 and 3.5 therefore because we're familiar with those let's go and compare the two looking at Google's most comprehensive model as of now they structure their pricing a little bit different than we see with open AI when it comes to input and output specifically they charge based off context and then you can kind of get discounted rates off less contextual window being used for

[02:04] reference you probably rarely would ever go over a 128k context unless you're putting in a ton of data so most of these numbers you can probably lean towards the 3.5 and the output of 10.5 or $3.50 and $10.50 just a real quick comparison that compared to open AI which has an input as of now of $5 per 1 million tokens and $15 per 1 million tokens so we're looking at a discounted price between $350 to5 and then $1050 to 15 knowing this depending on your output

[02:35] and depending on your workload you may choose between the two depending on which gives you the most reliable answers at scale coming over to 1.5 flash we can compare this pricing to maybe CAD gbt 3.5 with the input of. 35 we're looking at an input of 0.5 from Chad gbt and then coming down here for an output of $15 really random we got an output over here of $11.50 but pricing might not be everything you may want want to lean towards for quality but both the models could potentially provide you the value point that you're looking for therefore you may lean towards Gemini depending on

[03:07] your context we have gone ahead and got pricing out of the way one other really cool thing to keep in mind is that when you're just playing around with it getting comfortable with it it is free of charge also within Google AI Studio we have the ability to find two models and create prompts just to play around it's like almost like open AI playground that we saw earlier this is their version so when I create a prompt I can choose between a chat prompt and a structured prompt starting simple here let's go and just do a chat prompt and understand this user interface so right off the bat we can rename it we'll go ahead and say Chef Gemini not the Star

[03:39] Sign not talking about Scorpio Leo I'm talking about what Google decided to name their AI H say this is not barred anymore is not barred from here we have a couple other different options here such as the model so we saw that earlier we can choose between 1.0 pro flash Pro 1.5 Pro also if we create new tuned models we can choose those as well for this example here we're going to go with the 1.5 Pro it doesn't seem like you can set a Max tokens limit as of now but it has a max tokens count of like a million we're good there temperature what is temperature Corbin this is very valuable

[04:10] this is extremely valuable think of creating outputs at scale use your 333 gets the same type of output as user 967 this is where temperature comes into play a lower temperature is very consistent and follows your instructions to a T is really good for scaling and outputs a higher temperature is you're giving more creativity to the model typically when you're cre creating prompts you're going to go for low temp you're going to go very low like almost ice cold low like 0.1 low you also have the ability to adjust top P but in this context we only have to care about

[04:40] temperature they're kind of similar in regards to creativity meter so for now we're pretty set here you can ignore the stop sequence let's proceed now that we put in our variables let's understand the prompt itself so we have the ability to put in system instructions I want you to think of this as the context for the system so for example we can say give me precise and concise outputs for the type thing part here that's your user message that's going to be the data that we basically ask the chat bot to do or the API endpoint in this context so would be like how many pizza shops are in New York like questions like that now we're

[05:11] going to do a simple example together but if you want to see an in-depth video showing you everything you need to know when it comes to AI prompt engineering no matter the model check out that video right there go in depth show you how to create system instructions and show you how to create user instructions from a pretty pretty comprehensive perspective comprehensive perspective we'll let it slide to keep it simple here we're just going to say precise and concise outputs I didn't spell that right and then this will be the user message typically within your API call you'll be like messag quotation marks and provide if you're writing a Json and to be funny

[05:43] let's actually see if it can find how many pizza shops are in New York I highly doubt would be able to answer this how many pizza shops in New York New York hit run that's cool as a little bit of a ticker showing you how long it takes to calculate sadly it does not have access to that information I wasn't expecting it let's go and actually try a different question here so I'm going to go ahead and delete this and let's try a different question here so we're going to go ahead and say what is the population of the USA and there we go so we got our answer here 335 million notice how I say precise and concise outputs what happen if I change this let me go edit this real quick as well what

[06:13] happen if I change this to very long two paragraphs right I mean this just shows you system instructions are followed to a t I'm go and copy this delete these two and if I do again notice how the model is now giving very long answers comparative to that short concise answer that shows you the power of system instructions that is very very pertinent you set have good system instructions because a lot of time when you're proctoring with API endpoints most of your input uh for the like what to do is going to be the system instructions and your user message should just be the

[06:43] data you want to reformat restructure or do something with now with these outputs you can see there's a couple of different things we can do with them such as moving them up editing them or having a rerun click it you know give me a different answer let's see what it looks like if I keep rerunning the same prompt then obviously we can delete it copy rendered or copy marked down now a really big Point here that I like about this dashboard is the ability to see how many tokens are expended per run with this token counter here this can give you better context of how much it's going to cost you if you ran it like a thousand times check out that other video I referenced earlier if you want

[07:14] an in-depth tutorial on how to prompt engineer let's move on now Google AI studio also gives us the ability to create tuned models think of this like fine-tuning that we solve Chad gbt in open AI this is super cool I'm going to have to do a video on this make sure to subscribe as I'll do one in the future probably soon on how to leverage this this gives us the ability to upload a Google sheet CSV as context for the outputs and from there you can actually reference that as an API endpoint within your software so you don't have to worry about proctoring too much in system instructions because the actual

[07:44] underlying model itself understands the context did a whole video on this for fine tuning and Chad gbt model so we'll do one for Gemini as well seems like a big thing within this Google AI studio is the ability to integrate it within your drive as well so for context we can use sheets csvs images and other forms of dat within our Google AI Studio from here there is a ton of other resources you can look at when learning how to leverage this kind of technology I think a big one that you can check out is the prompt Gallery simply click it and we can go ahead and scroll down from all these different options here of different types of prompts that were created for a specific purpose so it's

[08:15] kind of like templates think of it that way for example we can do blog post Creator click it and it'll be directly added to our studio here with all of the relevant information filled out and I can simply hit run and here we go so we got the text image and and a little blog created about it now one important thing to do is make sure you are connected and logged in it'll prompt you throughout your workflow to connect your main account to this Google AI Studio it's going to allow you to save relevant chats from the past I'm going hit okay here and we can go ahead and reference them if we want to use them for the

[08:46] future now we have access to it as a nice little history tab that is a basic rundown of everything we can do within Google AI Studio as of now I'm going to make more like Niche videos on different topics about this studio so you can kind of get a better understanding of how to leverage Gemini's API especially in the context of software as now we're going to be able to create fine-tuned versions of Gemini's zi but also just understand how to prompt better in the context of software so make sure to leave a like it's completely free make sure to subscribe and I'll see you in the next video got two videos for you based off your clicks based off YouTube's outgo

[09:16] are they good are they bad I'll see you in the next video