OpenClaw

Use Fusion Model with OpenClaw.

Fusion configuration

Use a FUMO Dashboard key for an account registered and funded online.

Parameter	Value
Base URL	`https://capi.fumolab.ai/api/fusion/v1`
Available model IDs	`fusion-max-2.0`, `fusion-max-1.0`, `fusion-code-1.0`
API protocol	OpenAI Chat Completions API

What is OpenClaw?

OpenClaw↗ is an open-source agent platform that connects AI agents to messaging channels such as Telegram, Discord, Slack, Signal, iMessage, and WhatsApp.

Fusion connects as a custom provider through OpenClaw's openai-completions adapter.

Setup

Add Fusion as a custom provider in ~/.openclaw/openclaw.json.

Step 1: Get your FUMO API key

Follow Getting started for online self-service accounts (see Enterprise onboarding for enterprise accounts) to create and store your API key. If you use FUMO and have a sales representative, enter their Sales Code at signup or bind it later as described in Console and account.

Step 2: Set your API key

Export the key in the environment that starts the OpenClaw gateway:

export FUMO_API_KEY="your-key"

Add the export to your shell profile or production secret manager if the gateway must retain it after a restart.

Step 3: Configure Fusion

Merge the following provider and default-model settings into your configuration, preserving any existing settings:

{
  "agents": {
    "defaults": {
      "model": {
        "primary": "fumo/fusion-code-1.0"
      },
      "models": {
        "fumo/fusion-code-1.0": {
          "alias": "Fusion Code"
        },
        "fumo/fusion-max-2.0": {
          "alias": "Fusion Max 2.0"
        },
        "fumo/fusion-max-1.0": {
          "alias": "Fusion Max"
        }
      }
    }
  },
  "models": {
    "mode": "merge",
    "providers": {
      "fumo": {
        "baseUrl": "https://capi.fumolab.ai/api/fusion/v1",
        "apiKey": "${FUMO_API_KEY}",
        "api": "openai-completions",
        "models": [
          {
            "id": "fusion-code-1.0",
            "name": "Fusion Code",
            "reasoning": true,
            "input": ["text", "image"]
          },
          {
            "id": "fusion-max-2.0",
            "name": "Fusion Max 2.0",
            "reasoning": true,
            "input": ["text", "image"]
          },
          {
            "id": "fusion-max-1.0",
            "name": "Fusion Max",
            "reasoning": true,
            "input": ["text", "image"]
          }
        ]
      }
    }
  }
}

contextWindow and maxTokens are optional in OpenClaw. For longer workflows, use the context window and maximum output in Model specifications. For a service configured by sales, use the limits provided for that configuration.

Step 4: Start OpenClaw

After saving the configuration, start or restart the gateway:

openclaw gateway run

OpenClaw will now send the Fusion Code model label by default.

Model format

OpenClaw model references use provider/model:

fumo/fusion-code-1.0
fumo/fusion-max-1.0

The first fumo selects the custom provider. The remaining model ID is validated by the public gateway; after acceptance, the Fusion core treats it as a compatibility label and the service configuration associated with the key controls execution.

To send the Fusion Max model label by default, change agents.defaults.model.primary:

{
  "agents": {
    "defaults": {
      "model": {
        "primary": "fumo/fusion-max-1.0"
      }
    }
  }
}

Common errors

No API key found for provider `fumo`

Confirm that FUMO_API_KEY is exported in the gateway process environment, not only in a different terminal session. Restart the gateway after changing it.

Authentication error

Verify the API key you created and confirm that the ${FUMO_API_KEY} substitution has not been replaced with an empty value.

Model not working

Check that the model appears in both models.providers.fumo.models and agents.defaults.models, and that the primary reference includes the fumo/ provider prefix.

Requests use the wrong endpoint

This guide uses the openai-completions adapter. Fusion supports both Chat Completions and Responses, but OpenClaw applies adapter-specific request shaping. Keep the configured adapter and endpoint aligned.

Resources

Configure Hermes Agent

Supported models: fusion-max-2.0, fusion-max-1.0, fusion-code-1.0. Only Max 2.0 supports low / medium / high / max effort (default medium); the other two models do not support adjustment.