Zero‑to‑Hero OpenClaw Guide: Install, Configure, and Use the AI Gateway on Windows

Practice 1 – Local deployment on Windows 10

Prerequisites

OS: Windows 10 or Windows 11

Node.js 22 LTS (or Node 24, recommended)

Network access to GitHub and npm

Alibaba Cloud account for a Qwen API key

Step 1 – Obtain a Qwen API key

Open https://bailian.console.aliyun.com/ and log in.

Navigate to the API‑Key management page.

Create a new key; copy the value that looks like sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxx. Store it securely – it is shown only once.

Step 2 – Install Node.js

Download the LTS MSI from https://nodejs.org/ (e.g., node-v22.x-x64.msi, ~35 MB).

Run the installer and accept defaults.

Verify the installation:

C:\> node --version
v22.x.x
C:\> npm --version
10.x.x
✅ Installation successful

Step 3 – Install OpenClaw

Method 1 (recommended) – run the official PowerShell script: iwr -useb https://openclaw.ai/install.ps1 | iex The script checks for Node.js and npm, then installs OpenClaw. Expected output:

🔍 Checking system…
✅ Node.js 22.x installed
✅ npm 10.x installed
📦 Installing OpenClaw…
████████████████████ 100%
🎉 Installation complete!
Run <strong>openclaw init</strong> to start.

Method 2 – install via npm globally: npm install -g openclaw@latest If a sharp build error occurs, set the environment variable before reinstalling:

$env:SHARP_IGNORE_GLOBAL_LIBVIPS=1
npm install -g openclaw@latest

Step 4 – Configure the Qwen model

Run the onboarding wizard to create the configuration directory and enable auto‑start: openclaw onboard --install-daemon Edit C:\Users\<em>YourName</em>\.openclaw\openclaw.json (or use the CLI) to add the model and API key. Example JSON:

{
  "agents": {
    "defaults": {
      "model": {
        "primary": "dashscope-api/qwen-turbo",
        "fallbacks": ["dashscope-api/qwen-plus"]
      }
    }
  },
  "providers": {
    "dashscope-api": {
      "baseUrl": "https://dashscope.aliyuncs.com/compatible-mode/v1",
      "apiKey": "YOUR_API_KEY"
    }
  }
}

Alternatively, configure via CLI:

openclaw config set agents.defaults.model.primary "dashscope-api/qwen-turbo"
openclaw config set providers.dashscope-api.apiKey "sk-YOUR_API_KEY"
openclaw config set providers.dashscope-api.baseUrl "https://dashscope.aliyuncs.com/compatible-mode/v1"

Step 5 – Start the service

Run diagnostics: openclaw doctor If problems are reported, follow the suggested fixes.

Launch the gateway on port 18789: openclaw gateway --port 18789 Expected output:

✓ Gateway started on port 18789
✓ Control UI: http://127.0.0.1:18789

Open the control panel in a browser at http://127.0.0.1:18789. The UI shows the running status, selected model (Qwen Turbo), and channel configuration.

Step 6 – Test a conversation

In the UI, select the Chat tab.

Enter a message, e.g., “Hello, introduce yourself in one sentence.”

The AI replies, for example:

你好！我是 OpenClaw，一个可以帮你集成各种大模型的 AI 助理网关。有什么可以帮你的吗？

Common issues

Command not found: Ensure the npm global bin directory is in PATH.

Invalid API key: Verify the key was copied correctly and that the Alibaba Cloud account has sufficient quota.

Network timeout: Confirm connectivity to dashscope.aliyuncs.com and configure a proxy if needed.

Practice 2 – Configuring the Qwen large model

Model comparison (recommended for beginners)

qwen-turbo – fast, low cost; suitable for casual chat and simple tasks.

qwen-plus – balanced performance and cost; general‑purpose use.

qwen-max – strongest inference; best for complex tasks and code generation.

qwen-long – supports very long context; ideal for long‑document analysis.

Step 1 – Set the API key

Via configuration file (edit openclaw.json):

{
  "providers": {
    "dashscope-api": {
      "baseUrl": "https://dashscope.aliyuncs.com/compatible-mode/v1",
      "apiKey": "sk-YOUR_API_KEY"
    }
  }
}

Via CLI:

openclaw config set providers.dashscope-api.apiKey "sk-YOUR_API_KEY"

Step 2 – Add multiple models

{
  "agents": {
    "defaults": {
      "models": {
        "dashscope-api/qwen-turbo": {"alias": "Qwen Turbo"},
        "dashscope-api/qwen-plus": {"alias": "Qwen Plus"},
        "dashscope-api/qwen-max": {"alias": "Qwen Max"}
      }
    }
  }
}

During a chat you can switch models with the command /model qwen-max.

Step 3 – Test model invocation

openclaw message send --target "test" --message "你好"

Practice 3 – Basic usage tutorial

Step 1 – Open the control panel

Navigate to http://127.0.0.1:18789. The top menu contains Home , Chat , Skills , Sessions , and Settings . Status indicators show the service state, current model, and daily call count.

Step 2 – Send the first message

Click the Chat tab.

Type a message in the input box.

Press Send .

You: 你好
AI: 你好！有什么可以帮你的吗？

Step 3 – View conversation history

Select the Sessions tab and choose a session entry, e.g., “Session 1 – 2026‑03‑29 15:30”.

Step 4 – Adjust model parameters

In the Settings tab you can modify:

Temperature (default 0.7; higher values increase randomness).

Maximum token length (default 2048).

Click Save to apply changes.

Step 5 – Configure automatic session reset

{
  "session": {
    "reset": {
      "mode": "daily",
      "atHour": 4
    }
  }
}

Learning checklist

✅ Install OpenClaw independently.

✅ Configure a large‑model API.

✅ Chat via the Control UI.

✅ View session history.

✅ Adjust model parameters (temperature, max tokens).

✅ Set up daily session reset.