n8n Template: DeepSeek V3 Chat & R1 Reasoning Quick Start

Connect DeepSeek in n8n four ways: credentials → conversational Agent with memory → Ollama local → HTTP Request direct API

Who

Automation builders already running n8n who need to quickly wire up DeepSeek V3/R1

You get

A reusable workflow template with four integration paths; import, swap credentials, and run

You need

n8n · deepseek · ollama

Steps

1. Configure credentials and base parameters

   Create credentials in n8n: 1) openAiApi type, Base URL = https://api.deepseek.com, API Key from platform.deepseek.com/api_keys; 2) ollamaApi type pointing to your local Ollama service; 3) httpHeaderAuth type, Header name typically Authorization, value = Bearer + DeepSeek API Key. Sticky Note4 notes: you may also use https://api.deepseek.com/v1 as base_url; the v1 is unrelated to model version.

   Setup guidance · Source text。Template overview and quick-start guide

   Copy prompt

   This n8n workflow demonstrates multiple ways to harness DeepSeek's AI models in your automation pipeline! 🌟
   
   ## Core Features
   
   **Multiple Integration Methods** 🔌
   - Local deployment using Ollama for DeepSeek-R1
   - Direct API integration with DeepSeek Chat V3
   - Conversational agent with memory buffer
   - HTTP request implementation with both raw and JSON formats
   
   **Model Options** 🧠
   - DeepSeek Chat V3 for general conversation
   - DeepSeek-R1 for advanced reasoning
   - Memory-enabled agent for persistent context
   
   ## Quick Setup 🛠️
   
   **API Configuration**
   - Base URL: https://api.deepseek.com
   - Get your API key from platform.deepseek.com/api_keys
   
   **Local Setup** 💻
   - Install Ollama for local deployment
   - Set up DeepSeek-R1 via Ollama
   - Configure local credentials in n8n
   
   ## Implementation Details 🔧
   
   **Conversational Agent**
   - Window Buffer Memory for context
   - Customizable system messages
   - Built-in error handling with retries
   
   **API Endpoints** 🌐
   - Chat completions for V3 and R1 models
   - OpenAI API format compatibles

   Setup guidance · Source text。Sticky Note4 node content; credential configuration reference

   Copy prompt

   # Your First DeepSeek API Call
   
   The DeepSeek API uses an API format compatible with OpenAI. By modifying the configuration, you can use the OpenAI SDK or softwares compatible with the OpenAI API to access the DeepSeek API.
   
   https://api-docs.deepseek.com/
   
   ## Configuration Parameters
   
   | Parameter | Value |
   |-----------|--------|
   | base_url | https://api.deepseek.com |
   | api_key | https://platform.deepseek.com/api_keys |
   
   
   
   ## Important Notes
   
   - To be compatible with OpenAI, you can also use `https://api.deepseek.com/v1` as the base_url. Note that the v1 here has NO relationship with the model's version.
   
   - The deepseek-chat model has been upgraded to DeepSeek-V3. The API remains unchanged. You can invoke DeepSeek-V3 by specifying `model='deepseek-chat'`.
   
   - deepseek-reasoner is the latest reasoning model, DeepSeek-R1, released by DeepSeek. You can invoke DeepSeek-R1 by specifying `model='deepseek-reasoner'`.

2. Set up the conversational Agent with memory

   After importing the template, locate the AI Agent node (conversationalAgent) and Window Buffer Memory node. AI Agent systemMessage defaults to "You are a helpful assistant."; the DeepSeek node (lmChatOpenAi) has model = =deepseek-reasoner, credential = the openAiApi created in Step 1. Window Buffer Memory keeps default parameters; its ai_memory output connects to AI Agent's ai_memory input. When chat message received triggers, the Agent converses with DeepSeek-R1 using buffered context.

   Execution prompt · Source text。AI Agent node parameters.options.systemMessage

   Copy prompt

   You are a helpful assistant.

3. Configure the Ollama local model chain

   Locate Basic LLM Chain2 and Ollama DeepSeek nodes. Basic LLM Chain2 messages.messageValues[0].message is "You are a helpful assistant.". Ollama DeepSeek node (lmChatOllama) parameters: model=deepseek-r1:14b, options.format=default, options.numCtx=16384, options.temperature=0.6, credential=ollamaApi. Ollama DeepSeek's ai_languageModel output connects to Basic LLM Chain2's ai_languageModel input. When chat message received's main output connects to Basic LLM Chain2's main input.

   Execution prompt · Source text。Basic LLM Chain2 node parameters.messages.messageValues[0].message

   Copy prompt

   You are a helpful assistant.

   Execution prompt · Source text。Ollama DeepSeek node parameters.model

   Copy prompt

   deepseek-r1:14b

4. Configure HTTP Request direct API calls (JSON Body and Raw Body)

   The template provides two HTTP Request node examples. DeepSeek JSON Body node: URL=https://api.deepseek.com/chat/completions, Method=POST, specifyBody=json, jsonBody contains model=deepseek-chat, messages array (system role content = {{ $json.chatInput }}, user role = "Hello!"), stream=false, authentication=genericCredentialType, genericAuthType=httpHeaderAuth. DeepSeek Raw Body node: same URL, contentType=raw, rawContentType=application/json, body contains model=deepseek-reasoner, messages uses {{ $json.chatInput.trim() }}, stream=false, also using httpHeaderAuth. These correspond to direct V3 and R1 calls.

   Execution prompt · Source text。DeepSeek JSON Body node parameters.jsonBody

   Copy prompt

   ={
     "model": "deepseek-chat",
     "messages": [
       {
         "role": "system",
         "content": "{{ $json.chatInput }}"
       },
       {
         "role": "user",
         "content": "Hello!"
       }
     ],
     "stream": false
   }

   Execution prompt · Source text。DeepSeek Raw Body node parameters.body

   Copy prompt

   ={
           "model": "deepseek-reasoner",
           "messages": [
             {"role": "user", "content": "{{ $json.chatInput.trim() }}"}
           ],
           "stream": false
         }

Import this workflow template into n8n, then configure credentials, the Agent memory chain, the Ollama local node, and the HTTP Request direct nodes in order. After import, verify that ai_languageModel / ai_memory / main connections match the template, then replace all credential-id placeholders with live credentials.