Pular para o conteúdo principal

n8n

n8n can use Maritaca models through the OpenAI Chat Model sub-node. The configuration below was validated on n8n 2.30.7 with both a simple chain and an agent that executes a tool.

Prerequisites​

Keep the MARITACA_API_KEY value outside the workflow and only store it in an n8n credential.

Create the credential​

  1. Go to Settings > Credentials > Add Credential.
  2. Search for OpenAI.
  3. In API Key, paste the value of your MARITACA_API_KEY.
  4. Set Base URL to https://chat.maritaca.ai/api.
  5. Save it with a name such as Maritaca AI.

Configure OpenAI Chat Model​

Add OpenAI Chat Model to the chain or agent's Chat Model slot and set:

  • Credential: Maritaca AI.
  • Model: select ID mode and enter sabia-4 or sabia-4-thinking.
  • Use Responses API: disabled.

Disabling Use Responses API makes the sub-node use Chat Completions, which is the flow validated in the examples below.

Regular call with Basic LLM Chain​

Build this workflow:

Manual Trigger → Basic LLM Chain
                       │
                OpenAI Chat Model
  1. In Basic LLM Chain, select Define below for the prompt.
  2. Enter What is 25 + 27? Include 52 in the answer..
  3. Connect the configured OpenAI Chat Model.
  4. Execute the workflow and inspect the chain output.

Agent with Calculator​

Build this workflow:

Manual Trigger → AI Agent
                    ├── OpenAI Chat Model
                    └── Calculator
  1. In AI Agent, select Define below for the prompt.
  2. Enter Use the Calculator tool to calculate 6 times 7..
  3. Connect OpenAI Chat Model to the Chat Model slot.
  4. Add Calculator to the Tool slot.
  5. Execute the workflow; the final answer should include 42.

HTTP Request alternative​

For a direct request, configure an HTTP Request node:

  • Method: POST.
  • URL: https://chat.maritaca.ai/api/chat/completions.
  • Authentication: Header Auth.
  • Header Name: Authorization.
  • Header Value: Bearer YOUR_API_KEY.
  • Content-Type: application/json.
  • Body:
{
  "model": "sabia-4",
  "messages": [
    {
      "role": "user",
      "content": "What is 25 + 27? Include 52 in the answer."
    }
  ],
  "max_tokens": 100
}

To use the reasoning model, only change model to sabia-4-thinking. Do not put the API key directly in the workflow JSON.