1. OpenClaw Deployment

This chapter covers two parts:

• OpenClaw Installation
• Model Configuration (OpenRouter / Bailian)

1.1 OpenClaw Installation

The YahBoom version image already has OpenClaw installed (version 2026.4.24). Students using the YahBoom version image can directly skip to section 1.1.3

1.1.1 OpenClaw Installation Reference URL

OpenClaw Official Website: https://openclaw.ai/

Open Source Project Address: https://github.com/openclaw/openclaw

1.1.2 Using QuickStart Installation

We use the installation script provided by OpenClaw official:

​x
sudo apt install curl -y
curl -fsSL https://openclaw.ai/install.sh | bash
​
# Currently, the one-click download script installs the latest OpenClaw version by default. This version may not work well or be stable. If the installation fails or you get a permission error after installation, you can specify a version using the following command (reference version 2026.3.28)
curl -fsSL https://openclaw.ai/install.sh | bash -s -- --version 2026.3.28

During this process, the installation script will automatically detect missing dependencies and help us install them with one click.

During installation, use the arrow keys (up, down, left, right) to select options, spacebar to select/deselect, and Enter to confirm and configure.

Here we use the QuickStart option to quickly complete OpenClaw installation. Specific configurations can be done later.

If you haven't applied for any API Key, select Skip here. You can configure it later.

Here we keep the default model selection (configure later).

Select Channel here, mainly used for communication with third-party apps. We still skip it for now.

Search provider is the API provider for search services. We choose to skip.

Configure "Skills". We choose not to configure for now (select no).

Configure Hooks. We press spacebar to select "Skip for now", then press Enter to confirm.

Wait for installation to complete.

You need to add OpenClaw's installation location to the environment variable. Otherwise, you may not find the openclaw command in the terminal.

xxxxxxxxxx
echo 'export PATH="$(npm prefix -g)/bin:$PATH"' >> ~/.bashrc
source ~/.bashrc

1.1.3 Opening OpenClaw Control UI on PC (Windows)

1. Modify OpenClaw's configuration file to allow other devices on the LAN to access OpenClaw:

xxxxxxxxxx
openclaw config

Here we make selections as shown in the figure below (arrow keys to select, Enter to confirm):

Except for the Token field which can be different from the one in our image, everything else should be the same. The key point is the Gateway bind mode item. We need to set it to LAN to allow other devices on our local network to access our OpenClaw.

Finally, we select Continue to confirm and press Enter. Configuration is complete!

After completing the configuration, we first start the OpenClaw Gateway. Let's open another terminal and enter (after OpenClaw configuration is complete, the gateway may run automatically, depending on the installed version):

xxxxxxxxxx
openclaw gateway

2. Get Access Token

This Token is the one generated in our configuration. If you didn't record it, you can use:

xxxxxxxxxx
openclaw dashboard

This command displays the token once. Alternatively, you can find the ~/.openclaw/openclaw.json file. Your token will be configured in this file.

3. On another PC in the same local network, configure the environment to connect to S600.

We open the Cmd terminal on the computer (taking Windows PC as an example: press Win+R to wake up the quick RUN tool, enter cmd, then press Enter).

Enter in the console:

xxxxxxxxxx
ssh -L 18789:127.0.0.1:18789 sunrise@<sunrise-ip-address>

Note: Replace with the actual IP address of your RDK board.

After pressing Enter here, you need to enter a password (default: sunrise).

4. Access in Browser

We open the browser on the PC, and enter the IP+token string in the address bar (this is the address output by our openclaw dashboard just now, http://127.0.0.1:18789/#token=xxxxxxxxxxxxx). When this interface appears, deployment is successful!

1.2 Model Configuration (Must Read)

1.2.1 Why Configure Models

OpenClaw is an AI Agent gateway and does not have the "intelligence" capability of large models by itself. We need to configure large models to give OpenClaw the ability to think and understand.

1.2.2 How to Configure Models

Here we use three representative models as examples:

1. OpenRouter (built-in OpenClaw service provider, accessible both domestically and internationally)
2. Alibaba Bailian Large Model Platform (not an OpenClaw service provider, requires manual configuration)
3. Ollama (local large model)

Before starting configuration, we need to start the OpenClaw Gateway first.

If you did not exit the Gateway terminal after completing deployment in the previous chapter, you can skip this step.

We enter in the terminal:

xxxxxxxxxx
openclaw gateway

1. OpenRouter (Built-in OpenClaw Service Provider)

This requires you to have completed the OpenRouter API Key application according to the previous tutorial (please refer to the previous tutorial [Online Large Model / 4.OpenRouter Large Model API Aggregation Platform]).

The OpenRouter API Key format is like: sk-or-v1-b3exxxxxx4xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxcddf81d9

We still use the onboard wizard to configure:

xxxxxxxxxx
openclaw onboard

Please make selections according to the options shown in my figures.

Then enter the API Key you applied for and press Enter to confirm.

Next, we need to select a model. Due to OpenRouter's rate limiting policy, we recommend having more than $11 balance in your OpenRouter account. For insufficient balance, OpenRouter may throttle (this policy may change later, please refer to the latest actual situation). Here we can choose a free model (model names ending with free), or choose the OpenRouter free dedicated model openrouter/free (this is a router that randomly selects free models from available models on OpenRouter. This router intelligently filters models that support the features you requested).

At this point, you may be asked to fill in the API Key. We paste the API Key and press Enter to proceed to the next step.

Later, we come to other configuration items. Let's skip them for now and select Skip all.

At this point, we have completed the configuration.

We can use the TUI function to quickly test it.

Enter the command:

xxxxxxxxxx
openclaw tui

After completing the input, directly enter a greeting to openclaw (the Input part in the figure is the content we entered, and the green box is the model we selected).

2. Bailian (Third-party Service Provider API Example)

2.1 Apply for API Key

Before configuration, you need to apply for the API Key for Qwen. The application URL is:

https://modelstudio.console.alibabacloud.com/ap-southeast-1/?tab=doc#/doc/?type=model&url=2840914

After logging in, click the settings in the upper right corner and select:

Then click to create an API Key:

Click confirm:

As shown below, you have completed the API Key application:

Copy this API Key. Once closed, you cannot retrieve it again.

2.2 openclaw onboard Configuration

Enter in terminal:

xxxxxxxxxx
openclaw onboard

Refer to the figure below for selection:

After selecting Qwen, it will prompt to download related plugins:

Select this option to fill in the API Key you applied for:

After completing, select to show all model list:

Select the model you want to use:

So far, third-party model configuration is complete. For remaining operations, just select Skip. Then enter the following command in the terminal to restart the gateway:

xxxxxxxxxx
openclaw gateway restart

At first, it will show unknown. After some time, it will correctly identify the current model:

Similarly, besides the TUI interface, we can also chat and interact in OpenClaw's Control UI (the web version):

We can actually observe that the conversation content in TUI and the content in our Control UI are the same. This is because OpenClaw's conversation uses a Session mechanism to separate different sessions. Our TUI and Control UI are actually in the same Session. You can try clicking the New session button in the lower right corner to start a new conversation.