How to Run OpenClaw for Free with Kimi 2.5 (Full Setup)

Hey, I'm Dan. And in this video, I'm going to show you how to run OpenClaw for free with Kimmi 2.5. This will be a full setup, and I'll cover everything from start to finish. First, we'll set up the machine that you need to run OpenClaw. Then, we'll deploy OpenClaw, and finally, we'll connect a free messaging app so that you can use your AI assistant anytime. Before we get started, one thing that's really important to understand is that using OpenClaw usually involves two separate types of costs. The first type is what you pay to get your machine up and running. You want to use OpenClaw in an environment that's always online, and one that's protected from your personal machine for safety. The second type of cost is the ongoing cost of using AI credits with an API like OpenAI or Anthropic. These credits can really add up in a hurry, but at least for now, Kimmi 2.5 is free. Now that you understand the two types of costs, the first step is to set up a machine that we'll deploy OpenClaw to. Today, we'll deploy OpenClaw on a virtual private server. And a VPS is a good fit for OpenClaw because it offers safety benefits, scalability, and even though it isn't free, it's much cheaper than buying actual separate hardware like a Mac mini. We're going to use Hostinger to set up our VPS. Hostinger is very affordable to get started, and it makes deploying OpenClaw super simple with just one click. So, with a free AI model and an affordable hosting plan, this method is one of the most cost-effective ways to start using OpenClaw. And to get the best possible deal on your Hostinger VPS, make sure to sign up using our exclusive partner link in the description below to get an extra 10% off any Hostinger plan. Once you click on that link, you'll land on this page here. In the top right, you can select a VPS plan. KVM 2 is the default, and it's a good fit for OpenClaw. I'll keep it selected, and then I'll click deploy. Next, choose your billing period. I'll use 12 months for this tutorial, but 24 is usually the best value. Here you'll see that Open Claw is set to auto deploy, and that's exactly what we want. This just means that Hostinger will install it for you automatically. If you scroll down, you'll see a few more options. Because we're using Kimmy, we won't add any of these credits, and to save money, I won't enable daily auto backups. Though, if you're going to use this a lot, backups can be worth it. Next, pick a server location. Choosing one closest to your audience is usually best for performance, and my location looks to have the best latency. If you choose a 12 or 24-month period, and you've used our link, you'll see that the coupon code is automatically applied with the extra 10% discount. Once you do, and you're all ready, click continue. Then create your Hostinger account with an email and a password, and click register. On this screen, enter your billing information, then click continue to enter your payment information, and finally, when everything looks good, click submit payment. Once you've paid, click here to get started. Hostinger will bring you to the Open Claw configuration screen. Yours might look slightly different than mine, because Hostinger is continually improving this, but you'll see some fields where you could place API keys from major AI providers. Since we're using Kimmy, we won't enter anything in these, but if we scroll down, this Open Claw Gateway Token is a required field, and it's something that we need to save. Click here to make it visible, and then go ahead and copy this, and save it somewhere safe. This token is basically the key that lets your VPS connect to your Open Claw dashboard. So, treat it like a password. Don't share it, and don't leave it sitting in a doc that you'll forget about. Once you have it saved, you can rehide it, and then click deploy. You'll see a survey here, but you can just scroll down, and click skip. This will bring us to Hostinger's Docker Manager, and we can see that our project is being deployed. This may take a few minutes, so let's check back when it's finished. Once it finishes, your Open Claw project should show as running. Now that Open Claw is deployed, we need to connect and configure Kimmi 2.5. We haven't connected any models yet, and we need to do that before we can open our Open Claw dashboard and start chatting. To do that, click manage next to your Open Claw project. Then, with the visual editor selected, scroll down until you get to the environment section. Click on it to open, and scroll down to show all the names and value pairs. You'll see your Open Claw gateway token here. So, if you weren't able to save this when you first saw it, you can always come into these environment variables and click the I icon to show it. In this section, we need to add a new key. So, click the plus environment button here. And for the name, enter moonshot _api_key in all caps, just like I've done here. Then, we need to paste in a value. So, let me show you how we get this key. In a new tab, open the NVIDIA moonshot website. I'll make sure to put the link in the description below. Once you're on the site, you'll need to create a free account with NVIDIA. Over in the top right, click login. Then, enter your email and click next. Enter a password, complete the CAPTCHA, and create your account. Finally, click submit. Provide an account name, and then create an NVIDIA cloud account. The last thing you need to do to enable API access is to verify your account. Click on verify in the top right, and enter your phone number. Then, verify. Here, double-check that you're logged in, and then make sure Kimmi is selected in the top left. Then, click view code in the top right to generate an API key. In this window, click the generate API key button. Your new key starts with NVAPI found to the right of authorization. Including the NVAPI, copy this code and save it somewhere safe like you've done with your Open Claw Gateway token. Once you have it saved, you can close this window and then return to Hostinger. And then paste that same key into the value field. Once you have your Open Claw Gateway token and your Moonshot API key, scroll up and click deploy. Then we'll wait for your project to redeploy. If it spins for a while and prompts you to leave, that's okay. Click leave and then let's check to see if they've saved by clicking manage again. Scroll down and open the environment tab and we can see that they've saved. Now you can scroll back up and cancel to get back. Now that you've added your Moonshot API key, let's configure Open Claw and test the chat to make sure everything is working. Click the drop-down arrow here and then click this link to open your Open Claw dashboard in a new tab. Enter your Open Claw Gateway token and hit login. You'll see a message saying that no AI provider API keys are configured. That's totally fine. We're going to set up Kimmy manually in just a second. For now, just click continue and I'll walk you through the rest of the setup step-by-step. Once Open Claw opens, you'll land in the chat tab. We'll come back to this shortly once everything has been configured. First, click the overview tab in the top left. Here you can double-check that your gateway token has been entered and that the status shows as connected. Next, back on the left and under settings, click config. Then select raw to open the config editor. This is where we tell Open Claw which model to use. And we're going to point it to Kimmy K 2.5 and pass in the API key that we just added. And don't worry, if you're not used to JSON, I know that this part looks a little intimidating. You don't need to understand everything in here. If you've set the environment variables correctly from the previous step and you use the JSON that I've provided in the description, then it should be pretty easy. So, open the link to the JSON from the description, copy it, then go back to OpenClaw, select everything that exists in the JSON currently, remove it, and then paste in the JSON that we copied from the file. It now tells OpenClaw to use Kimmy K 2.5 as our model. You can see that here. And as long as you've named your variables correctly, you'll see those variables moonshot_api_key and you'll see openclaw_gateway_token in this JSON as well. If you have those correctly, this setup should work for you with your keys without needing to paste those keys in directly. Click save and update. At this point, your health should still look okay. If not, and you're having trouble with this, it can be really helpful to troubleshoot with the help of another language model. For instance, if you ever need to make changes to the JSON config in the future, sharing screenshots or code or error messages can help you troubleshoot even if you only have access to something like OpenAI. Now that your config settings are updated, we can test to see that we're connected to the Kimmy K 2.5 API. Scroll back up to the chat, select it, and then we're going to send a simple test message. I'll ask, "Does it seem like you're correctly using Kimmy K 2.5 with the NVIDIA API?" And then, I'll click send. Here you can see your assistant typing, which is a good sign. And if your assistant responds, then Open Claw is running on your VPS and your Kimmy API key is working. If you don't get a response, you can check the health in the top right, and you can go back to the overview to check that your status is connected. And again, if you're having issues, try troubleshooting with another AI that you do have access to already. Take note that the Nvidia free Kimmy API is relatively slow. If it's too slow for what you're working on or testing, one option is to go directly to the Moonshot developer console. You can pay for Kimmy 2.5 and you can set it up using a similar process with your new key. Although you'll be paying for credits, it is a much cheaper option than OpenAI and Anthropic, and Kimmy still seems pretty impressive. With your VPS set up, Open Claw deployed, and Kimmy 2.5 working, the last step is to connect a messaging app so that you can actually chat with your assistant like a normal conversation. This is one of the best parts of using Open Claw. You can chat with your assistant right from your phone or computer, no dashboard necessary. For this tutorial, we'll be using Telegram. It's the easiest option since it supports bots and tokens, and it's a quick way to confirm everything is working. Let me show you how to set it up. In the Open Claw chat, I'll type, "Please help me set up Telegram." And your Open Claw AI assistant will respond with steps. Yours might look a little different from mine, but the main steps will be the same. I'll be using Telegram for desktop in this tutorial because it's easiest to set up, and the main thing it's going to ask for is a Telegram bot token, and we'll get that from the Bot Father. So, let's open Telegram and search for this Bot Father. I'll search @BotFather, and you want to find this verified one and click it. In the Bot Father conversation, click start. If you scroll up, you can see the full list of commands, and you can either type out new bot or just click the command here. Here you need to pick a name for your bot and I'll call mine MM for Metics Media Open Claw Kimmi. Send. Next, it asks for a username. This one has to be unique and it has to end with bot. So, I'll try MM Open Claw Kimmi bot. Once it creates the bot, Bot Father will first give you a token. Treat this like you've treated your other API keys. Go ahead and copy it. Go back to Open Claw and tell it that you have the token that Bot Father gave you from step five. Paste it into the chat and then click send. This may take a minute to configure. Once it's done, you just need to pair your new bot to your Open Claw setup. Back in Bot Father, click the bot to open it. Then hit start. It should respond with your Telegram user ID and your pairing code. I'll just copy both lines here. Copy the selected text. Go back to Open Claw and provide them in the chat. And once Open Claw approves your pairing, you can go back to Telegram and we could try to send a quick test message. I'll say, "Hey, are you online?" And you'll see that your bot is typing and if it replies, you're all connected. If you go back to Open Claw once more, you should see that same message in your Open Claw chat. And that's all you need to do to use Kimmi as a free service. You now have Open Claw deployed on a VPS, connected to your Nvidia Kimmi API, and linked to Telegram so you can message it from anywhere. If you haven't signed up for a hosting or VPS yet, remember to use the first link in the description. That link applies our exclusive discount. It gives you an extra 10% off any hosting or plan and it makes this one of the most affordable ways to get started. Thanks for watching.