OpenClaw/ Hermes Configuration

OpenClaw and Hermes support custom OpenAI-compatible services that can be accessed through MoonNexAI Base URL to chat, visual understanding and other compatible API.

Configure Values

Configure Item	Value
API Type	OpenAI-compatible / VLLM
Base URL	https://moonnexai.com/v1
API Key	<MOONNEXAI_API_KEY>
Chat Endpoint	/chat/completions
Common Chat Model	gpt-5.6-terra, or model ID returned using GET /v1/models

Ready, API Key.

1. Login https://moonnexai.com.
2. Go to console, confirm account is operational.
3. Create or copy API Key on the API token page.
4. Save it to its own safe location and do not enter the project warehouse.

Run Configuration Command

Run at terminal:

openclaw config

Windows can use CMD or PowerShell; MacOS can use Terminal.

When you select a configuration location, select Local (this machine).

Select Model in the configuration entry.

Select the VLLM or OpenAI-compatible type.

Fill in by hint:

VLLM base URL: https://moonnexai.com/v1
VLLM API key: <MOONNEXAI_API_KEY>
VLLM model: gpt-5.6-terra

If you want to use another model, call GET /v1/models and fill in the id of the response to the model name position.

Example request

If the client configuration fails after the client, you can first confirm the API Key and the model with the curl:

curl https://moonnexai.com/v1/chat/completions \
  -H "Authorization: Bearer <MOONNEXAI_API_KEY>" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.6-terra",
    "messages": [
      {
        "role": "user",
        "content": "用一句话介绍 MoonNexAI。"
      }
    ]
  }'

Visual input

The client that supports visual input usually puts the picture in messages[].content. The configuration still uses the same Base URL:

{
  "model": "gpt-5.6-terra",
  "messages": [
    {
      "role": "user",
      "content": [
        {
          "type": "text",
          "text": "请描述这张图片。"
        },
        {
          "type": "image_url",
          "image_url": {
            "url": "https://example.com/image.png"
          }
        }
      ]
    }
  ]
}

Check.

phenomena	Treatment
Connection failed	Confirms the Base URL as https://moonnexai.com/v1.
Return authentication Error	Confirms that API Key does not have any additional spaces and Bearer is used for request head.
Return Model Error	Replaces the model name that GET /v1/models returns.
Picture unrecognized	Confirms that the model supports visual input, and that the picture URL is directly accessible.

More generic settings are available at Client configuration.