OpenClaw has rapidly emerged as a leading consumer-grade platform for AI agent orchestration, seamlessly connecting your digital life — calendars, emails, and Notion workspaces — into fully automated workflows. However, an AI agent is only as powerful as the model and inference infrastructure driving it.
While OpenClaw handles your agent tools and workflow automation, DGrid delivers the intelligence. By integrating DGrid, you bypass the heavy local hardware requirements of running on-device models (like Ollama), while avoiding the high latency and costly usage fees of legacy LLM providers. DGrid’s high-performance API gives you instant access to 200+ leading models (including OpenAI GPT and Google Gemini) with enterprise-grade speed and reliability.
Unlike existing setup guides that require manual JSON file editing and command-line configuration, this tutorial uses visual, no-code setup via OpenClaw’s native Dashboard Control UI. Anyone can complete this end-to-end setup, regardless of coding experience.
Prerequisites
Before you begin, ensure you have the following ready:
- Terminal Access:
Terminal(for macOS/Linux) or Administrator PowerShell (for Windows) - Node.js 22 or Higher: OpenClaw’s core functionality requires Node.js v22+.
- Check your current version by running
node -vin your terminal/PowerShell - Update to the latest LTS version if your install is older than v22 before proceeding
- Check your current version by running
- Active DGrid Account & API Key: You will need a DGrid API key to connect the service. We will cover how to get this in Step 1.
Step 1: Get Your DGrid API Key
First, generate your secure DGrid API key to authenticate the connection between OpenClaw and DGrid’s inference infrastructure:
- Go to the official DGrid Console and connect your wallet (or sign up for a free account if you are new)
- On the credits page, select Keys
- Click Create New Key, give it a recognizable name (e.g., “OpenClaw”)
- (Optional but recommended) Enter the credit limit and/or expiration date.
- Copy the generated API key (starts with
sk-) and store it in a secure location
Critical Note: Never share or publicly expose your API key, as it grants access to your DGrid account and usage.
Step 2: OpenClaw Installation
Install the full OpenClaw core on your machine with a single, copy-paste command. No code editing or configuration is needed at this stage.
For macOS & Linux Users
- Open your Terminal app
- Copy and paste the following command, then press Enter:
curl -fsSL https://openclaw.ai/install.sh | bash - Wait for the installation to complete (this typically takes 1–2 minutes).
For Windows Users
- Open PowerShell as Administrator (right-click PowerShell > “Run as Administrator”)
- Copy and paste the following command, then press Enter:
iwr -useb https://openclaw.ai/install.ps1 | iex - Wait for the installation to complete (this typically takes 1–2 minutes).
Once the installation finishes, you will see a success message in your terminal/PowerShell confirming OpenClaw is installed on your system.
Step 3: OpenClaw Initialization & No-Config Onboarding
In this step, we will initialize OpenClaw and complete the onboarding wizard with all model configuration skipped. This avoids any manual file editing, and we will set up DGrid entirely via the visual dashboard in the next step.
- In the same terminal/PowerShell window, run the following command to start the onboarding process and install the OpenClaw background daemon:
openclaw onboard --install-daemon
Select ‘Yes’ to agree to the terms of use:
- The onboarding wizard will launch in your terminal. Follow these exact selections for a no-code setup:
| Wizard Prompt | Required Selection | Reasoning |
|---|---|---|
| Onboard mode | QuickStart | Uses the streamlined, default setup flow for fast onboarding |
| Config handling | Use existing values | Retains default system settings (we will customize the model provider via the UI later) |
| Model/auth Provider | Skip for now | Critical: Skips all command-line model configuration, so we can set up DGrid visually |
| Filter models by provider | No selection needed / All providers | No action required, as we skipped provider setup |
| Default model | Keep Current | Retains the default placeholder, which we will update to our DGrid model in the dashboard |
| Select channel(Quick Start) | Skip for now | Default is pairing; unknown DMs get a paring code |
| Plugins (Skills) installation | Skip for now | Ensures a stable core setup before adding extra complexity |
| Enable hooks | Skip for now | Press space to select, enter to submit |
Restart the service after gateway is installed and select Hatch in TUI:
- Once you complete the wizard, OpenClaw will finish initializing the daemon. Open the OpenClaw Dashboard Control UI in your default browser.
- If the dashboard does not open automatically, you can also manually navigate to the local dashboard at:
http://127.0.0.1:18789/orhttp://localhost:18789/
- If the dashboard does not open automatically, you can also manually navigate to the local dashboard at:
Dashboard Authentication (If Prompted)
If the UI asks for an authentication token:
- Return to your terminal/PowerShell and run:
openclaw config get gateway.auth.token - Copy the generated token from the output
- Paste the token into the auth field in the dashboard settings, then click Connect.
- If you encounter issues, generate a new token with:
openclaw doctor --generate-gateway-token
Step 4: Configure DGrid AI RPC Service via Dashboard Control UI
This is the core no-code setup step. We will add DGrid as a model provider, configure your API credentials, and add supported models — all via point-and-click in the OpenClaw Control UI, with zero file editing.
4.1 Navigate to Model Configuration Settings
- In the OpenClaw Dashboard Control UI, locate the right-hand sidebar menu
- Click Settings, then select Config from the dropdown
- In the Config menu, select the Models > Models Providers tab. This is where we will add our DGrid provider and models.
4.2 Add DGrid as a New Model Provider
- In the Providers section of the Models tab, click the Add Entry button to open the configuration form
- Fill in the following fields exactly as shown below (these values are optimized for DGrid’s API):
| Field Name | Value to Enter |
|---|---|
| Name the Entry | DGrid |
| Model Provider API Adapter | OpenAI Responses |
| Model Provider API Key | Your full DGrid API Key (copied in Step 1, starts withsk-) |
| Model Provider Auth Mode | api-key |
| Model Provider Authorization Header | True |
| Model Provider Base URL | https://api.dgrid.ai/api/v1 |
4.3 Add Models to Your Provider
Now we will add the specific AI models you want to use from DGrid’s library of 200+ models. Below are the two most popular, high-performance models (you can add more later using the same steps):
Add GPT-5.3-Codex(Recommended Default)
- Click the Add button in the Model Provider Model List section
- Fill in the model configuration fields as follows:
| Field Name | Value to Enter |
|---|---|
| API | OpenAI Responses |
| Context Window | 400,000 |
| Model ID | openai/gpt-5.3-codex |
| Max Tokens | 128,000 |
| Model Name | GPT-5.3 Codex |
- Click Save to add the model to your DGrid provider.
Add Claude Opus 4.6
- Again, under the Model Provider Model List section, click Add
- Fill in the model configuration fields as follows:
| Field Name | Value to Enter |
|---|---|
| API | OpenAI Responses |
| Context Window | 200,000 |
| Model ID | anthropic/claude-opus-4-6 |
| Max Tokens | 128,000 |
| Model Name | Claude Opus 4.6 |
- Click Save to add the model.
Pro Tip: You can repeat this process to add any of the 200+ models supported by DGrid’s API. Just update the Model ID, Name, Context Window, and Max Tokens to match the model specifications.
4.4 Set DGrid as Your Default Primary Model
To ensure your OpenClaw agents use DGrid’s models by default for all tasks:
- In the left-hand Settings menu, navigate to Agents > Defaults
- For the Primary Model field, enter:
dgrid/openai/gpt-5.3-codex
Format Note: Use the lowercase provider name (
dgrid) followed by the full Model ID, separated by a forward slash
- Click Save to apply your global default model settings.
All configuration changes are now live, with zero code edits required.
Step 5: Verify Your DGrid & OpenClaw Integration
To confirm your setup is working correctly and OpenClaw is connected to DGrid’s API:
- Return to the main Chat tab in the OpenClaw Dashboard Control UI
- Send a simple test query, such as:
“Hello! Please confirm which model you are running on, and which API provider you are connected to.”
- Check the response: OpenClaw will confirm it is using the DGrid provider and your selected default model (GPT-5.3 Codex)
- For additional validation, send a more complex reasoning or creative task to confirm low latency and full model functionality.
If you receive a valid response, your setup is complete! You now have a fully functional OpenClaw instance powered by DGrid’s high-performance AI inference API.
Troubleshooting Common Issues
“Unauthorized” / Error 1008
- Ensure the OpenClaw daemon is running: run
openclaw statusin your terminal to check - Verify you are using the correct gateway auth token (re-generate with
openclaw doctor --generate-gateway-tokenif needed) - Confirm you are accessing the dashboard via
http://127.0.0.1:18789/on your local machine
Models Not Showing Up / Connection Failed
- Double-check your DGrid provider configuration: ensure the Base URL and API Key are entered correctly with no extra spaces
- Confirm your DGrid API key is active and has available usage credits in the DGrid Dashboard
- Verify the Model ID matches the exact ID supported by DGrid’s API
- Save all configuration changes and refresh the dashboard to reload settings
Support
If you encounter any issues with your DGrid API key, model access, or integration, please visit the DGrid Dashboard or contact our support team for assistance.
Thank you for choosing DGrid. Happy automating!