Claude Code + Ollama: FREE Local AI Coding FOREVER (Step-by-Step Tutorial)

Hi guys, welcome back. And today in this

video be talking about how you can use

cloud code without paying even a single

dollar. So we are going to be using the

power of cloud code with our local large

language model running within our own

machine. And you can see that the

pricing of the cloud code at the moment

is like $1.17 if you are going to be

going with an annual subscription

discount and $1200 will be charged up

front while you're going to be

purchasing this particular plan. And

this is only in US dollar which is

mentioned. But if I'm buying in New

Zealand, it is $34 if I'm going to buy

this particular subscription over here,

which is to be honest quite huge. And if

I'm going to be spending this much of

money, I know that it's going to be very

very faster because cloud code is super

fast and the models that they have is

super powerful. But what if we have an

ability to run everything within our

local machine without spending a single

dollar and that we can achieve with the

power of the cloud code itself, which

they have got the access to all lama

models these days. So if you just go to

the cloud code over here as you can see

over here this is the cloud code that we

always know and we have used before and

the installation is quite

straightforward. This is how you install

it. If you really not installed it

before I'll quickly show you in this

particular video like how you can do it

and once you do the installation then if

you just go to Olama over here they now

have an ability to go and connect with

the cloud code and use the models which

are there in the Olama models itself. So

you can basically use the models from

Olama within your cloud code itself. And

this time I'm going to be using my Asus

GX10 super computer which is this one as

I have already shown demo before and

this particular computer as you can see

come with a Nvidia Blackwell chip and it

is like 128 GB of uh the GPU memory as

well and it's it's really really really

very very powerful. So I'm going to be

using this particular machine for this

demonstration while I'm going to show

you. I can also use my Apple M1 machine

as well, but it is bit slower

comparatively the Asus uh GX10 machine

itself. So, I'm going to be using the

Asus GX10 and I'm going to show you how

it's amazingly you can use everything

from the ground up. Well, as said, let's

get started in the video and I will show

you everything and I'm quite excited to

show you what I have got installed

today.

[music]

So, the first thing we need to do is to

install the cloud code. So, the first

thing is I'm going to go copy this

particular command as you can see over

here. So I am actually in the Olama docs

itself and in the integration you have

something called as coding and in the

coding you have something called as

cloud code. So you can just use the

cloud code over there and I am just

going to see in the documentation they

have mentioned that the cloud code

through Olama anthropy compatible API

enables to use the model such as GLM4.7

QN3 coder and GPT OSS model. So these

are the recommended model by the uh by

the Olama team itself to use with the

Olama model that you have got and uh I

think I'm just going to go with what

they have told. So I'm going to show you

how amazing it they are. So I'm going to

first do the installation of the cloud

code. So I'm going to go copy this and

I'm going to open my terminal over here

and I'm going to paste this command. I'm

going to hit enter. There we go. The

installation is done. It also tells you

that the native installation exist but

the uh local bin is not in your path. So

you need to add it. So I need to either

add in the z file or in the bash rc

file. But I'm just going to do the

export command for now instead of just

doing it for this demonstration. So I'm

going to go paste this over here. So now

I have the cloud code uh running. So I'm

just going to do a source of the zshrc

which means it's going to just reload

the terminal session. Uh and now if I'm

going to do a cloud over here, it is

just going to work. See if it is green

color, which means it's it's happy. And

now it's uh going to work for us over

there. So I'm going to go uh with one of

the project that I have got which is

going to be the EA um EA app uh project

which is this one. And I have opened

this particular application in the Ryder

IDE as you can see over here. And if you

wanted to use the writer IDE and all the

products from JetBrines for free of cost

for 3 months, you can use the link in

the description below which is going to

give you the uh the offer for 3 months

for free of cost. So if you just go to

the website over here, let's say

jetbrians.com

uh and you can see that they have got

the uh the coding agents natively

integrated in the ide. So you can use

that particular part over there free of

cost as well. And if you just go to the

all the products over here. So all these

products that you are seeing over here

are available for free of cost for 3

months. Just use the coupon code below.

It is going to give you the discount.

Thank you JetBrian for making this

happen. Well, as that said, this is the

rider ID that I'm going to be using for

this particular demonstration over here.

And if I'm going to run this particular

application, I quickly show you how this

application looks like. So if I'm going

to run this app, it is going to

basically open two um two pages. one is

the web URL and another one is the back

end uh with an API over here. So if I'm

going to just run this particular

product, it is the product that you are

seeing over here. The list of product

that I have got. I can either create a

product or I can either delete the

product or edit the product or I can

view the product. So all of these things

I can do from here itself. It's a very

very super simple UI that I have got and

I have also got the APIs and if you have

watched my Udemy courses before you know

how this application is built and how

this application is tested in playright

selenium with AI I have also covered

everything with this core with this

particular application. I have did many

time with this particular application.

So this is not a new if you have already

following my Udemy courses. So I'm going

to just go to the hyper terminal over

here one more time and because we have

already installed the cloud code uh now

we need to use the lama to make it

happen. So the way you can actually do

it is there is a command called as lama

and then there is a command called as

launch. So you need to make sure that

you have updated the latest version of

Olama as well. If not these features may

not work. uh just update the Olama to

the latest version and [snorts] then see

that there is something called as claude

over here and then just use the command

hyphen config and if you hit enter it is

going to show you all the models which

are running within my ASUS GX10 machine

and how do I connected this I have

already talked about that in my other

video but I'll quickly show you how I

did it I am using the uh Nvidia spark

link option which does the connectivity

for me over here see this is the way

that I have connected to my ASUS GX10

machine. Uh this is the Olama that I'm

connected with. And you can see that

just if I'm going to show you one more

time over here in a different window

over here. So if I'm going to just do an

Oola Lama list, you can see that I have

got these many models over here. So

these are the models which are there

within my ASUS GX10. But the moment I'm

going to disconnect with my uh Asus GX10

and if I'm going to be doing an Olama

list this time these are the model which

are running within my local machine. So

the model is see they are completely

different. There is a Kim K2.5 cloud

model GP2 uh GPT OSS 120 billion

parameter deepsee uh v1 3.1 671 billion

cloud model. So these are the model

which are running within my local

machine as you can see over here. uh and

the moment I'm going to connect with the

Asus GX10 over here using the Nvidia

sync and if I'm going to connect to the

Olama over there. Now what happens is

the models are going to be different as

you can see see that these are the

models are coming from the ASUS GX10

itself and you can see that the the

parameter that I'm running it over here

120 billion parameter which is 65GB of

storage uh it's taking and 20 billion

parameter and there is a Q uh three

quarter 30 billion parameter as well. So

I'm going to use this particular model

this time and I'm going to see how it

works. So I'm going to copy this

particular model. I'm going to close

this window because it's not required.

And because I've already connected to it

now see that the moment I'm going to do

lama launch cloud of config it is going

to show me all the model that is there

within my machine. See that all these

models which are there in my local Olama

of ASUS GX10 and I'm going to choose

this uh QN3 quarter 30 billion parameter

and the moment I select it it's going to

say do you want me to launch the uh

cloud code now? And I'm going to say

yes. And the moment I'm going to say yes

over here, now all the configurations

are going to be delegated to cloud code

from this point on. Yes. See launching

cloud code with Q3 quarter 30 billion

parameters. So now it is connected to

cloud code. So if I'm going to say /mod

uh which is this one and you can see

that it shows me that there are these

models available but the one that I'm

connected to is the QN3 quarter 30

billion which is a custom model. Wow.

which is cool. So we have connected to

the local uh model over here. So I'm

going to just say escape. And now I'm

going to ask it to write some code. So

the code which I'm going to ask it to

write itself is that uh you can see that

the UI was quite normal, right? Like

it's not really modern or anything like

that. But I wanted this UI to be

modernized. Uh and I wanted the the UI

to look and feel more modern approach

instead of having it like this. So I'm

going to say uh maybe I'm going to say

this EA web app, right?

Can you try to modernize the UI of my EA

web app application which is built using

C.NET

MVC

framework. That's all. And now I'm going

to just hit enter. So from this point on

it is going to start doing the app

buildings and everything within my local

large language model over here. Uh

instead of using anything from the cloud

itself. So I don't even have the cloud

subscription right now. I have

completely got rid of it and I'm

actually using everything from my local

large language model itself. And this

machine I know it's very very powerful

machine. It's also very costly but at

least you see that this is the way that

you can actually use these kinds of

machine for doing the development

purpose as well as testing purpose using

the local models and there are many

advantages of having these machines uh

and also having these uh cloud over here

because see that the the token is not

that bad. It is doing quite good at the

end of this particular uh execution that

you are seeing over here. I will show

you how many tokens been generated or

used to perform this operation and we'll

see how it's going to look like. So I'm

just going to wait for the entire

execution to happen. Ah look at that. It

has already found that the UI this is

what it is and we want to modernize

things. So there is an UIUX improvement.

Uh it says migrate the bootstrap 5 for

better responsiveness and modern

component uh and blah blah blah and

component modernization technology/

stacks and the performance improvement

accessibility. Wow, that's pretty cool.

I didn't even know any of these. So I'm

going to just say uh yes, allow all the

edit during the session giving the

entire um stuff to my local large

language model. And because this is

local large language model uh and of

course I have control over this model. I

can unplug this anytime. Uh now I'm just

going to see how things are going to

work. So let's just wait for this to

happen.

All right, you can see that the changes

have been implemented right now. So we

have got everything done over here with

all the changes that this tool was doing

all these. I'm just scrolling up over

here. You can see that how many changes

it did around 23,000 tokens it was it

was actually doing to make this happen

to complete this entire task. So now let

me go and stop this entire execution and

probably rebuild the solution because

this is a net code. So I need to rebuild

it. Uh oop I can see there is an error

coming up over here somewhere. If I'm

just going to go up a bit uh there is a

red color line there. Oh possible

conflict of the assert with the same

target. Uh there we go. I think we have

some error here. So the build has got

some error. So I'm going to ask the same

to my cloud code and we'll see how that

works. Okay, there we go. And now I have

asked the cloud code to uh to see if

there is any um any way to fix this

particular issue. So I'm just uh ask

that. So let's wait for uh this to be

resolved. So it's again running it. Now

I will need to wait and see how much

time it's going to take. So last time

the entire application building was

taking around 15 minutes. It's it's not

as uh faster as you can imagine. It is

slower. Uh if you can do the same thing

with the with the cloud models, for

example, cloud code 4.5 or Opus 4.6,

it's going to be super super faster. But

this is this is slower to be honest

because it's all running in my local

machine and it's just getting warmer as

well uh to be honest. And uh I will just

need to wait and see how long the entire

fix is going to take. Yeah, it's doing

something. So uh let's see how long it's

going to complete. So I'll just wait for

the fix to be fully uh done and then

I'll be back. All right, finally the

error is also gone because I just

executed the check the error if it is

gone and it is seems to be gone and

there is this particular UI as you can

see. Finally, it has built this

particular UI as you can see. It creates

a product and the UI is completely

amazing. Uh, and there is also 24

products been listed which is also been

shown over here. And there is a

homepage. Look at that. Like how amazing

it is. And there is a product uh page

over here. Also shows the type as

peripheral something like that. And you

can see uh the view and there's an edit

button. Wow, this is pretty cool. So all

of these are happening just from our

local large language model as you can

imagine and this is working as expected.

This is the power of the local large

language model running on the uh ASUS

GX10 uh and also how you can work

everything offline instead of going on

uh to the internet by using the powerful

large language model something like

that. So yeah, I can see that it's all

just working fine and it's just working

as expected. And I could see that this

there is a potential of using the local

large language models running on the

local machine with the cloud code. But

if you ask me if they are very faster, I

would probably say no. They're not even

close to the the models that are running

on the cloud. If you have a cloud model,

they do way more faster than running on

the local machine. That's what I can

see. And also uh they are not quite

reliable as you can do it with the cloud

models. That's my honest opinion. But

but still if you think that you have

some use cases to use your local large

language model and do all these

development, you can still get these

kind of operations. But it's going to

take a long time. I did this entire

recording for more than 25 minutes to

get this part. But if I'm going to do

the exact same thing with the cloud

model, I could have done that in less

than five or 6 minutes. That's the max.

That's it guys. Once again, thank you so

much for watching this video. This is

how you can use cloud code with Olama

and you can use the local larger

language model to do all of these

operation. Thank you so much. Catch you

in the next one.