Using n8n

To connect n8n to Tutujin API, edit an OpenAI credential: use your Tutujin key as the API key and change the Base URL from OpenAI's default to https://www.tutujin.com/v1. Every OpenAI-related node using that credential then sends requests to Tutujin API, and model names are the ones in the model hub.

Step 1: Create a credential

1. Open Credentials and create a new OpenAI credential.
2. Fill in the following and save:

When you save, n8n fetches the model list with this credential; if it saves, the URL and key are correct.

Step 2: Use it in a workflow

In AI workflows (AI Agent, Basic LLM Chain and so on), attach an OpenAI Chat Model sub-node:

1. Set Credential to the credential from step 1.
2. Pick Model from the list, which shows the models this key can use, or switch to entering an ID and type a model name such as deepseek-v4.1-flash.
3. Set the other options (temperature, max tokens…) as needed.

To embed text, for example when writing documents to a vector store, use the Embeddings OpenAI node with the same credential and an embedding model name such as text-embedding-3-small.

Calling the API with HTTP Request

If you would rather not use AI nodes, or need an endpoint they do not cover, use the HTTP Request node:

{
  "model": "deepseek-v4.1-flash",
  "messages": [
    { "role": "system", "content": "Classify the customer feedback as shipping, quality, after-sales or other. Output only the category." },
    { "role": "user", "content": "My parcel still hasn't arrived after three days" }
  ]
}

The reply is in choices[0].message.content of the result, ready for later nodes. For other endpoints, change the URL and body; parameters are in the API reference.

Keeping costs under control

Scheduled and batch workflows in n8n can send many requests in a short time. Give n8n its own key with a spending limit, and control batch sizes in loop nodes to avoid rate limits — see Rate limits. The cost of every call is listed under Usage in the console.

Troubleshooting