Resources · n8n integration

n8n Cohere Chat Model nodeConfigure Cohere Chat Model in n8n.

The n8n Cohere Chat Model node never runs on its own. It is a sub-node, a small node you hang off a bigger one: no main input, and a single output that plugs into the Model port of an AI Agent or a chain. Two parameters to set, a model and its options, and the root node talks to Cohere.

Try the Cohere Chat Model node on n8nSee the 2 parameters

Verified Trustpilot reviews · AI, automation & growth agency

Why automate

What does the n8n Cohere Chat Model node actually do?

It supplies the language model that another node uses to generate text. In n8n, an AI root node such as AI Agent, Basic LLM Chain or Summarization Chain receives the workflow items through its main input and does the reasoning, but it holds no model of its own. You attach one through a port, a small connector under the root node. This node fills the port named Model with Cohere.

Three shapes cover most of what builders do with it. The first one is a conversational assistant: an AI Agent handles the loop, the Cohere Chat Model answers each turn, and the agent calls tools that read a Google Sheets file or send a Slack message. The second is a batch of text jobs, a chain that rewrites or classifies rows one item at a time, where Sampling Temperature pushed down keeps the output stable enough to write back into a spreadsheet. The third is retrieval: a question and answer chain reads documents and the model turns the retrieved passages into an answer.

The node is also a way to swap providers without touching the rest of the workflow. Because every chat model sub-node exposes the same port type, unplugging OpenAI or Anthropic and plugging Cohere in its place leaves the root node, its prompt and its tools untouched. Two public templates show the pattern, one that generates and evaluates prompts across several providers at once (workflow 16063), and one vision retrieval workflow built on Cohere models (workflow 6961).

When to reach for something else. If the root node is an agent that needs to pick tools and you are only after one completion on one item, a simpler chain does the job with the same sub-node attached. If you want to call the Cohere API for something the chat interface does not cover, this node is not the right door: it exists to feed a Model port, not to expose the whole API surface.

The limits are worth knowing before you build. The node ships at version 1 and exposes 2 parameters, the model selector and an options collection holding 2 fields, so anything beyond temperature and retries is not configurable here. It has no main input, which means you cannot run it alone to see what it returns: you test it by running the root node it is wired to. And the model list is loaded from your Cohere account, so a credential that cannot reach the API gives you an empty dropdown rather than an error at execution time.

Cohere bills the API calls it receives under its own terms, and n8n adds nothing on top of that. The node itself costs nothing, on n8n Cloud and on a self-hosted instance alike. If you are weighing platforms before committing, the n8n review goes through hosting and pricing models in detail.

Connect

What do you need to connect before it runs?

  1. 01

    Create the Cohere credential

    Open the Credentials menu in n8n and add a Cohere credential. n8n stores it once and every workflow on the instance can reuse it, so you do not paste the same secret into each node. The official n8n documentation for this node points to its Cohere credential page for the exact fields to fill.

  2. 02

    Select it in the node

    Drop the Cohere Chat Model node on the canvas and pick the credential you just created in the credential selector at the top of its panel. There is no Authentication dropdown on this node, so nothing else to choose: one credential, and the panel is ready for the Model parameter below it.

  3. 03

    Open the model list

    Click the Model field. n8n queries your Cohere account and fills the list with the models it returns, which doubles as a connection test: a populated dropdown means the credential works. An empty list points back at the credential rather than at the workflow.

Connections

Where does this node plug in?

n8n splits its AI nodes in two: root nodes take the workflow items through their main input and do the work, and sub-nodes hang off their ports, one type of sub-node per port. This one has no input of its own, only an output that goes up into a root node.

Output (what it plugs into)

  • Modelai_languageModel
01ModelThe output port of the node, and its only connection. Drag it onto the Model connector under a root node and that node routes every generation through Cohere.

The output port of the node, and its only connection. Drag it onto the Model connector under a root node and that node routes every generation through Cohere.

Key parameters

  • Required: without this link the node is inert, since it has no main input of its own.
  • AI Agent and AI Agent Tool: agent root nodes, for a model that reasons and calls tools in a loop.
  • Basic LLM Chain and Question and Answer Chain: plain chains, for one completion per item or an answer grounded in retrieved documents.
  • Text Classifier, Sentiment Analysis and Information Extractor: task nodes that need a model to sort, score or pull fields out of text.
Use cases
a support workflow where the same Cohere model serves an agent and a Text Classifier that tags incoming messages.
Parameters

Which parameters do you set?

The Cohere Chat Model node has 2 parameters. For each one: the node as you configure it in n8n, what the parameter changes, and our field notes.

01

Model

model

What you see in n8n

Notes & use cases

Picks the Cohere model that generates the completion. n8n loads the list dynamically from the Cohere API with your credential, so the choices you see are the ones your account can actually call.

Key parameters

  • Model: a dropdown fed by Cohere at the moment you open it, not a fixed list baked into the node.
Use cases
two branches of the same workflow can run two copies of this node on different models, one for the drafting step and one for the classification step, both wired to their own root node.
02

Options

options

What you see in n8n

Notes & use cases

A collection, meaning a field that stays empty until you add entries through Add Option. Leave it closed and Cohere applies its own defaults; open it when a run comes back too creative or too fragile.

Key parameters

  • Sampling Temperature: controls randomness. A higher value creates more diverse sampling but raises the risk of hallucinations; as it approaches zero the model turns deterministic and repetitive.
  • Max Retries: how many times n8n retries a failed request before giving up and failing the root node.
Use cases
a nightly job that rewrites product descriptions into a sheet runs at low temperature, so the same row does not come back reworded every night.
Need help

Need help automating Cohere Chat Model with n8n?

A person reads every message.

FAQ

Questions builders ask next

01Is the Cohere Chat Model node included in n8n?
Yes. It belongs to the n8n AI package, which ships with n8n itself, on n8n Cloud and on a self-hosted instance alike. There is nothing to install, no community node to add, and no extra charge on the n8n side. The Community Edition under the Sustainable Use license includes it. What is not free is Cohere: the provider bills the API calls your workflows send under its own terms, and n8n adds nothing on top. So the cost of a workflow that uses this node is whatever your model usage costs at Cohere, billed by Cohere, on your own account.
02What do you need for the n8n Cohere Chat Model node to work?
Two things. A Cohere credential, created once from the Credentials menu and reusable across every workflow on the instance, and a root node to plug into. The node has no Authentication selector, so the credential is the only thing to pick in that part of the panel. On the connection side, the output goes into the Model port of an AI Agent, a chain such as Basic LLM Chain, or another root node that accepts a language model. Without that link, the node sits on the canvas doing nothing at all.
03What are the limits of this node?
It exposes 2 parameters at version 1: the model selector and an options collection with 2 fields, Sampling Temperature and Max Retries. Anything else you might want to steer is not surfaced here. It has no main input, so it cannot be executed alone and cannot be tested in isolation, and it cannot receive items from an earlier node directly. The model dropdown depends on a working credential, since n8n reads the list from the Cohere API rather than hardcoding it. It is also limited to chat generation, not to the rest of what Cohere offers.
04What do you have to connect at minimum, and which sub-node to start with?
At minimum, one root node and one credential. Start with Basic LLM Chain rather than AI Agent: the chain takes an item, sends one prompt, returns one completion, and there is only one moving part to debug. Once that works, swap the chain for an AI Agent and add memory and tools on the other ports. If you are comparing providers, the sub-nodes for OpenAI, Anthropic or Google Gemini fill the same Model port, so the rest of the workflow does not change when you unplug one and plug another.
05n8n or Make for this kind of AI workflow?
It depends on where you want the workflow to live and how you want to pay for it. n8n can be self-hosted with Docker or npm, or used on n8n Cloud, and the workflow is the same either way, which matters when the prompts carry data you would rather keep on your own infrastructure. Make is hosted only, with no self-hosting option, and is billed per operation, which is easier to start with and easier to predict on low volume. Both are visual. Pick on hosting, data control and cost model rather than on the node list.
Hack'celeration Lab

Get our weekly integration tips.

No spam. Unsubscribe anytime.

Same platform

Other docs you might need on this stack.

Browse all integrations