AI agents and workflows are the buzzwords in 2025, and with each Agent, lies a set of AI models that enable these agents function, reason and run. There are different models provided by different companies eg OpenAI provides the chatGPT models, Anthropic Provides the Claude models, Meta has Ollama etc
Each of these offers different functional and reasoning capabilities affecting how you can use them either as tools or chat completion functions
What is DeepSeek
DeepSeek is a Chinese AI model that is quickly gaining momentum. it offers a chatGPT4 like experience for a fraction of the cost. This makes it more suitable and favorable for use in agents and tools within the AI community
What is n8n
N8n is a fair code used for business automations and building AI agents. Its offers more advantages as its low code and integrates with most models easily. They offer both the paid cloud version and a free self hosted version that you can run on your PC or cloud servers
How to use DeepSeek in N8N
To use DeepSeek models in n8n, follow the following steps
- Install the community DeepSeek node and use it for chat completions
- Use it for Tool calling purposes
Methos 1: Installing the n8n community node
This is only applicable to the self hosted users
- Go to n8n settings, and select the community nodes option
2. Click on the install button. You should see a popup
3. Go to this link to copy the npm package name as n8n-nodes-deepseek .
4. Copy and paste the exact name on the n8n community node popup, and check the box and install
5. Create a new workflow and enjoy the new DeepSeek node
Method 2: Using DeepSeek in N8N Tools
This is the more practical method to use DeepSeek with n8n agents and tools
- Add an AI agent to your workflow
2. On the chat Model, add the Open AI model
3. Create a new Open AI credential with your DeepSeek API. This will give you an error but don’t worry
4. On the Open AI node, under options, click on the option for base URL
5. Change the base URL to https://api.deepseek.com/ and refresh the model being used
6. Test it out. Add a simple tool and test whether the tool works. If you get an error try debugging the steps above, or checking the Model being used
Questions and ideas?
Feel free to email me at Zacharia@effibotics.com