Run OpenClaw Locally with Ollama (100% Private and Free)

[00:01] Right now, there are two big reasons why

[00:03] people are hesitating to use open cloud.

[00:06] Number one, it is not free. You're

[00:08] paying for every token, every task,

[00:11] every agent use. That's money out of

[00:13] your pocket.

[00:15] And number two, every time you send

[00:16] something to open cloud, it goes out on

[00:19] someone else's server. Whether it's

[00:21] ChatGPT or Anthropic, your data leaves

[00:24] your computer, your contacts, personal

[00:25] projects, all of that goes out on the

[00:28] internet.

[00:29] What if I tell you there's a way to use

[00:31] open cloud without

[00:33] using any token and your data stays on

[00:36] your computer? So, today, let me

[00:38] introduce you how to use Ollama to power

[00:42] your open cloud with a local model. This

[00:45] is completely free and your data is 100%

[00:47] safe. So, how does this actually work?

[00:51] There are two pieces. The first piece is

[00:53] Ollama.

[00:54] Think of it as a platform that runs AI

[00:57] models locally on your computer. It

[01:00] downloads the model, hosts it, and keeps

[01:02] it running in the background. The second

[01:04] piece, obviously, is open cloud. So,

[01:06] normally, open cloud goes out to a

[01:09] provider like Anthropic for [music] its

[01:11] intelligence.

[01:12] But, when you point it at Ollama

[01:14] instead,

[01:16] it calls for local model for

[01:17] intelligence. This way, no internet is

[01:20] involved, your data is completely safe.

[01:23] So, the very first thing to do is just

[01:25] to download Ollama. Head down to the

[01:27] Ollama official website,

[01:29] ollama.com/download,

[01:31] and uh download the installation file.

[01:33] And after doing that, just install

[01:35] Ollama. Okay, now we open up this

[01:38] application, Ollama, and uh we have a

[01:41] fun task of picking a model which your

[01:44] open cloud will be powered on.

[01:46] So, just go to Ollama

[01:48] official page, and these are a list of

[01:51] models you can download to your

[01:53] computer,

[01:54] which can then be used to power open

[01:56] cloud. But, before you pick a model,

[01:58] there's one thing you need to

[01:59] understand.

[02:00] The most important thing about the model

[02:02] is the parameters. The more parameter a

[02:05] model has, the smarter it is. But, the

[02:07] trade-off is it also takes up more space

[02:10] on your computer and needs more

[02:11] computing power. So, if you have a

[02:13] regular laptop, you probably want to

[02:16] start with something small.

[02:18] For this tutorial, I'm going to pick up

[02:20] a very popular model that performs

[02:23] relatively well. It's called Kimmy K

[02:25] 2.5. Click on this copy button and then

[02:29] open Open up a terminal,

[02:32] paste in [music] this command, run it.

[02:36] It will say something like, "If your

[02:38] browser did not open, navigate to this."

[02:40] So, copy this URL,

[02:43] go back to your browser

[02:45] >> [music]

[02:45] >> and paste it.

[02:47] Connect.

[02:49] Go back to terminal, paste in this

[02:51] command again, and hit [music] enter.

[02:54] Connecting to Kimmy 2.5. Send a message.

[02:56] Just say hi.

[03:00] Okay, so this means your Kimmy model has

[03:03] been installed on your computer. Go to

[03:06] Ollama.

[03:07] You should be able to find that model,

[03:09] Kimmy K 2.5.

[03:12] The next thing we want to do is to

[03:13] install and launch open cloud via

[03:16] Ollama.

[03:17] To do it, also very straightforward,

[03:20] head down to this link

[03:22] and copy this command, Ollama launch

[03:25] open cloud.

[03:27] Copy it, go back to terminal,

[03:30] paste it, and hit enter.

[03:34] Open cloud is not installed. Install

[03:35] with npm. Say yes. Uh it will ask you

[03:38] which model do you want to use to power

[03:40] open cloud. So, here, just press enter

[03:43] on Kimmy 2.5.

[03:46] Proceed. Yes.

[03:48] And I understand the risk. Continue.

[03:50] Yes.

[03:51] So, if you're seeing this, that means

[03:54] your open cloud has been successfully

[03:57] installed. And sometimes, if you get hit

[03:59] with an error message, that means your

[04:01] node version is not up to date. What you

[04:05] need to do is to just update node. And

[04:08] the command to do that is

[04:11] nvm install node.

[04:13] So, right now, your open cloud is up and

[04:15] running. If you want to see the

[04:16] dashboard, scroll to this section. This

[04:20] is the URL for your open cloud

[04:22] dashboard. Copy the URL,

[04:25] go to a browser, paste it.

[04:29] And here you go. You can configure your

[04:31] channel, your skills, and everything in

[04:35] here.

[04:36] You want If you want to test it out, you

[04:37] can simply go back to the terminal and

[04:39] say, "Who are you?"

[04:44] Because we're running this on a local

[04:45] model,

[04:46] it's going to take a lot of time for it

[04:48] to answer your questions.

[04:50] So, to be honest with you, local models

[04:52] are definitely not as powerful as

[04:55] ChatGPT or cloud.

[04:57] If you give it a complex, multi-step

[04:59] task, it will struggle. The answers are

[05:02] generally shorter, less nuanced, and

[05:04] sometimes just worse. But, this is free

[05:07] and uh it keeps your data safe. So, make

[05:10] that trade-off and see if you want to

[05:12] use open cloud this way.