Resources · n8n integration

n8n Mistral Cloud Chat Model nodeConfigure Mistral Cloud Chat Model in n8n.

The n8n Mistral Cloud Chat Model node is the brain you clip onto an agent, not a step in your workflow. It exposes 7 parameters, from the model picker down to Random Seed, and it plugs into a single port called Model. Built for anyone wiring a Mistral model into an n8n automation.

Try the Mistral Cloud Chat Model node on n8nSee the 7 parameters

Verified Trustpilot reviews · AI, automation & growth agency

Why automate

What does the n8n Mistral Cloud Chat Model node actually do?

It hands a Mistral chat model to a root node so that root node can think. On its own it produces nothing: it has no main input, it never runs alone, and it appears on the canvas as a small block hanging under an agent or a chain. A sub-node, in n8n language, is exactly that: a component a bigger node borrows through a port, which is the little connector drawn under the root node.

The clearest scenario is a support answer written for you. A chat trigger feeds an AI Agent, the agent reads the question, and this node supplies the completion. You pick the model in Model, cap the answer with Maximum Number of Tokens so a chatty reply never turns into a wall of text, and leave everything else alone. Two parameters, one working assistant.

Second scenario: a classification pass that has to give the same verdict twice. Feed tickets into a chain, set Sampling Temperature low and fill Random Seed, and repeated calls stop drifting between runs. That matters the day someone reruns yesterday's execution and expects the same labels. Route the result into Slack and the team sees a stable answer, not a new opinion each morning.

Third scenario: a public-facing generator. Enable Safe Mode injects a safety prompt before every conversation, which is the switch you flip when the output lands somewhere customers read it. Pair it with a narrow Top P and the model stops reaching for improbable phrasing.

When to reach for something else. This node only speaks to Mistral Cloud. If your account is on another provider, swap it for the OpenAI, Google Gemini or Anthropic chat model node instead: same port, same wiring, different credential. And if the root node you want is an agent exposed as a tool to another agent, that is AI Agent Tool, not a different model node.

The limits are worth knowing before you build. The node has no memory of its own, no tools, no output parser: those are separate sub-nodes on separate ports of the root node. The model list is loaded from Mistral Cloud and shows only what your account can reach, so a colleague's screenshot may not match yours. And n8n charges nothing for the node itself, while Mistral bills its own API calls under its own terms.

Connect

How do you connect the node to Mistral Cloud?

  1. 01

    Create the credential once

    Open the node and use the credential selector at the top of its panel, or go through the Credentials menu in n8n and create the Mistral Cloud entry there. A credential is a saved set of authentication details that lives outside the workflow, so every future workflow reuses the same one instead of holding a copy of the key.

  2. 02

    Attach it to the node

    Back in the node panel, select the credential you just saved. Nothing else on the node asks for authentication: there is no method selector, no scope list, no per-workflow setting. One credential attached, and the node is allowed to call Mistral Cloud on behalf of your account.

  3. 03

    Open the Model list to prove it works

    Click Model. n8n loads the list live from Mistral Cloud, which makes it the fastest credential test in the panel: names appear, the connection is good. Pick one, then drag the node's connector onto the Model port of your agent or chain so the root node knows which brain to use.

Connections

What do you plug this node into?

n8n splits AI work in two: a root node holds the logic and receives the workflow items, and sub-nodes clip onto its ports to supply a brain, a memory or a tool. This node is a sub-node with one output port and no main input.

Output (what it plugs into)

  • Modelai_languageModel
01ModelThis is the single port the node exposes, and the one the root node reads from when it needs a completion. Drag the connector from the bottom of the node up to the Model slot drawn under the root node, and the pairing is done.

This is the single port the node exposes, and the one the root node reads from when it needs a completion. Drag the connector from the bottom of the node up to the Model slot drawn under the root node, and the pairing is done.

Key parameters

  • Required: the root node cannot run without a model on this port.
  • AI Agent: the usual root, and the one to try first when the workflow has to hold a conversation.
  • Basic LLM Chain: the leaner root, one prompt in and one answer out.
  • Text Classifier: a reading root, for sorting messages instead of replying to them.
Use cases
one Mistral node can serve an agent handling a chat flow, while a second one, set colder, serves the Text Classifier that sorts the same messages.
Parameters

Which parameters of the n8n node matter?

The Mistral Cloud Chat Model node has 7 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 Mistral model that generates the completion. It is the only parameter outside the options list, and the only one you always touch.

Key parameters

  • Model: a dropdown loaded live from Mistral Cloud, showing the models your account is allowed to call.
Use cases
a summarizer that runs on every inbound email can sit on a small model, while the agent answering customers gets a larger one. Same workflow, two nodes, two entries in this dropdown.
02

Maximum Number of Tokens

options.maxTokens

What you see in n8n

Notes & use cases

Caps how long the completion can get. A token is a chunk of text, roughly a short word, and the ceiling here is about the answer, not about what you send in.

Key parameters

  • Maximum Number of Tokens: the maximum number of tokens generated in the completion. Most models carry a context length of 2048 tokens, and the newest ones go well beyond that.
Use cases
a node writing a one-line ticket subject gets a tight ceiling so the model stops instead of adding a paragraph of politeness nobody reads.
03

Sampling Temperature

options.temperature

What you see in n8n

Notes & use cases

Governs how adventurous the sampling gets. Lower it and the completions become less random; push it toward zero and the model turns deterministic and repetitive.

Key parameters

  • Sampling Temperature: a number controlling randomness, typically driven straight from the panel rather than from an expression.
Use cases
a node that extracts an invoice reference wants the boring answer every time, so it runs near the floor. A node drafting subject lines has room to breathe higher up the scale.
04

Max Retries

options.maxRetries

What you see in n8n

Notes & use cases

Decides how many times n8n tries the request again before the node gives up and the execution fails.

Key parameters

  • Max Retries: the maximum number of retries to attempt for a single request.
Use cases
on a nightly batch that nobody watches, a couple of retries absorb a hiccup on the provider side. On a chat flow where a human is waiting, keeping it low means the workflow fails fast instead of leaving someone staring at a spinner.
05

Top P

options.topP

What you see in n8n

Notes & use cases

Narrows the pool of words the model is allowed to pick from, through nucleus sampling. At 0.5, half of all likelihood-weighted options are considered.

Key parameters

  • Top P: a number setting that pool. The general advice is to alter this or Sampling Temperature, not both at once.
Use cases
a product description generator that keeps wandering into odd vocabulary calms down with a lower value here, while temperature stays untouched so the two settings never fight each other.
06

Enable Safe Mode

options.safeMode

What you see in n8n

Notes & use cases

Switches on a safety prompt that gets injected before every conversation the node handles.

Key parameters

  • Enable Safe Mode: a checkbox. On, the safety prompt goes in ahead of the exchange; off, it does not.
Use cases
any node whose output is published without a human reading it first, such as an automated reply on a public channel, deserves this box ticked. An internal node summarizing your own database rarely needs it.
07

Random Seed

options.randomSeed

What you see in n8n

Notes & use cases

Fixes the starting point of random sampling. With a seed set, separate calls generate deterministic results instead of a fresh variation each time.

Key parameters

  • Random Seed: the seed used for random sampling, left empty unless you specifically want repeatable output.
Use cases
testing a prompt change is painful when the answer moves on its own. Set a seed while you iterate, compare two runs honestly, then clear it before the workflow goes live if you want variety back.
Need help

Need help automating Mistral Cloud Chat Model with n8n?

A person reads every message.

FAQ

Questions people ask next

01Is the n8n Mistral Cloud Chat Model node free to use?
Yes. It ships inside the n8n AI package, so there is nothing to install and nothing extra to pay on the n8n side, on n8n Cloud as well as on a self-hosted instance running the Community Edition under the Sustainable Use licence. A workflow built in one place behaves the same in the other. The one cost that is not n8n's is the model itself: Mistral bills the API calls this node makes according to its own terms, exactly as it would if you called the API by hand. n8n adds nothing on top of that.
02What do you need for the node to work?
Two things. A Mistral Cloud credential, created once from the node panel or the Credentials menu and reused across every workflow afterwards, and a root node to hang it on. The node has no main input and never executes by itself, so an agent, a chain or another root node has to claim it through the Model port. There is no authentication selector on the node, no scope to tick, no per-workflow setup: attach the credential, open the model list, connect the port, and it runs.
03What are the limits of this node?
It supplies a chat model and nothing else. Memory, tools and output parsers are separate sub-nodes that clip onto other ports of the root node, so this one cannot remember a previous message or call an API on its own. Version 1 exposes 7 parameters: Model plus 6 options. Anything Mistral offers that is not on that list is out of reach here. The model dropdown is filled from your own account, so it can differ from a colleague's screen, and an older workflow may still display an earlier node version with fewer options.
04What do you have to connect, and which sub-node should you start with?
At minimum, one chat model on the Model port of a root node. Start with an AI Agent as the root: it is the most forgiving, it accepts this node directly, and you can add memory or tools later without rebuilding anything. Basic LLM Chain is the leaner option when you only want one prompt in and one answer out, with no decisions to make. Information Extractor, Text Classifier and Sentiment Analysis also accept this node when the job is reading text rather than conversing with someone.
05n8n or Make for running a Mistral model?
It depends on where your data should live and how you want to pay. n8n can be self-hosted, which means the prompts and the model responses stay on infrastructure you control, and the pricing model is not tied to how many operations a workflow burns. Make is hosted only, billed per operation, and its visual editor is often quicker to pick up for a linear flow. For an AI workflow that loops, branches and calls several sub-nodes, the n8n canvas shows the wiring more literally, which is what you want when you are debugging a model that answers oddly.
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