n8n has a built-in DeepSeek Chat Model node, but it is a sub-node: it plugs into an AI Agent or another AI root node as the model, and isn’t a step you can use on its own. Community nodes fill two gaps around it: a standalone DeepSeek node for plain workflow steps, and chat model variants that handle DeepSeek’s reasoning and thinking features differently. This guide covers the built-in node and its credential (from n8n’s documentation) and the most used DeepSeek community packages (from our own npm data and the packages’ READMEs).
Quick comparison
| Built-in DeepSeek Chat Model | DeepSeek community nodes | |
|---|---|---|
| Type | Sub-node: the model for an AI Agent or other AI root node | Either a standalone node or a chat model sub-node, depending on the package |
| Credential | DeepSeek API key | Usually a DeepSeek API key (see each README) |
| Comes with n8n | Yes | No, installed as a community package |
| n8n Cloud | Available | Only packages that n8n has verified |
Step 1: get a DeepSeek API key
- Log in to DeepSeek or create an account.
- Open the API keys page.
- Select Create new secret key and, if you like, name it.
- Copy the key into the n8n DeepSeek credential as the API Key.
The built-in DeepSeek Chat Model node
Use it to give DeepSeek’s chat models to a conversational agent. Its one parameter is Model: n8n loads the list from DeepSeek, so you only see the models your account can use. Options:
- Base URL: override the default API URL.
- Frequency Penalty: higher values make the model less likely to repeat itself.
- Maximum Number of Tokens: sets the completion length.
- Response Format: Text or JSON. JSON makes the model return valid JSON.
- Presence Penalty: higher values make the model more likely to talk about new topics.
- Sampling Temperature: higher values give more varied output, with a higher risk of hallucinations.
- Timeout (milliseconds) and Max Retries.
- Top P: lower values ignore less probable options.
n8n notes that DeepSeek’s API is compatible with OpenAI’s, and points to LangChain’s OpenAI documentation for more detail.
DeepSeek community nodes
Our DeepSeek hub lists 28 packages that pass our quality rule, with 89,376 npm downloads a week between them. The most downloaded, with the description each author gives on npm (weekly npm downloads as of 27 October 2026):
| Package | Weekly downloads | Author’s description |
|---|---|---|
| n8n-nodes-deepseek | 88,048 | A DeepSeek AI node, similar to OpenAI |
| n8n-nodes-deepseek-v4-thinking-fix | 175 | DeepSeek V4 models (deepseek-v4-flash, deepseek-v4-pro) with thinking mode control |
| @gtvh2o/n8n-nodes-deepseek-fix | 166 | DeepSeek V4 models with thinking mode control |
| n8n-nodes-deepseek-reasoner | 137 | DeepSeek R1/Reasoner Chat Model with reasoning_content handling for tool calls |
| n8n-nodes-deepseek-fix | 131 | DeepSeek V4 models with thinking mode control |
| n8n-nodes-deepseek-chat-model | 123 | DeepSeek Chat Model with reasoning content preservation |
| n8n-nodes-deepseek-llm | 65 | A DeepSeek AI node, similar to OpenAI |
| n8n-nodes-deepseek-ai-chat | 60 | DeepSeek AI Chat Model sub-node for AI Agent workflows (V4-native) |
| n8n-nodes-deepseek-thinking | 47 | Exposes DeepSeek’s thinking level (reasoning_effort from none to max) |
n8n-nodes-deepseek: a standalone node
This package has nearly all of the hub’s downloads. According to its README, it adds a DeepSeek node that you use as a normal workflow step: add it, pick the model and enter your messages, with an API key credential. That makes it the option when you want a DeepSeek completion without building an AI Agent.
Reasoning and thinking variants
Most other packages are chat model sub-nodes that change how DeepSeek’s reasoning output is handled. The README of n8n-nodes-deepseek-reasoner, for example, says it fixes a “Missing reasoning_content field” error that its author saw when using DeepSeek R1 with an AI Agent and tools, because the built-in node’s underlying LangChain model drops the reasoning_content field between tool-calling turns. Its instructions: use the AI Agent as usual and connect DeepSeek Reasoner Chat Model as the Chat Model sub-node instead of the built-in one. Several other packages (the V4 “thinking” and “fix” packages, deepseek-chat-model, deepseek-thinking) describe thinking mode or reasoning content control in the same way. Read each README to see which models and settings it supports, and check its node page here for the latest release and, where it has one, its GitHub repository for open issues.
How to choose
- DeepSeek as the brain of an AI Agent: start with the built-in DeepSeek Chat Model.
- An agent with tools that fails on reasoning content, or you need thinking mode settings: try one of the reasoning or thinking sub-node packages and compare their READMEs.
- A plain DeepSeek step without an agent: n8n-nodes-deepseek. Since the API is OpenAI-compatible, n8n’s own docs also point to the OpenAI-style setup for details.
To install a package on a self-hosted instance, an owner or admin enters its npm name in Settings > Community Nodes; full steps are in How to Install n8n Community Nodes. The settings that control community packages are listed in n8n Community Node Environment Variables. On n8n Cloud only verified community nodes can be installed, see Verified vs Unverified n8n Community Nodes. For connecting agents to outside tools, see MCP in n8n.
FAQ
Does n8n support DeepSeek without community nodes?
Yes, through the built-in DeepSeek Chat Model sub-node, which gives DeepSeek’s models to an AI Agent. It uses a DeepSeek API key.
Can I use DeepSeek in n8n without an AI Agent?
Not with the built-in node, which is a sub-node. The community package n8n-nodes-deepseek adds a standalone DeepSeek node.
Which DeepSeek models can I pick in n8n?
The built-in node loads the list from DeepSeek, so it shows the models available to your account.
What is the most used DeepSeek community node?
n8n-nodes-deepseek, with 88,048 weekly npm downloads on 27 October 2026.
