Resources · n8n integration

n8n Qwen Cloud Chat Model nodeConfigure Qwen Cloud Chat Model in n8n.

This node is an engine you plug into another AI node, never a step that runs on its own. It sends chat prompts to a model hosted on Qwen Cloud and exposes 9 parameters, from model choice to sampling, timeout and retries. Written for builders wiring an n8n Qwen Cloud Chat Model node into a chain.

Try the Qwen Cloud Chat Model node on n8nSee the 9 parameters

Verified Trustpilot reviews · AI, automation & growth agency

Why automate

What does the n8n Qwen Cloud Chat Model node do?

It supplies the language model that another node uses to produce text. n8n splits its AI nodes in two: a root node such as AI Agent receives the workflow items through its main input, and sub-nodes attach to its ports, one kind of sub-node per port. This node is a sub-node on the model port. It holds the Qwen Cloud credential, the model name and the sampling settings, then hands answers back to whatever asked for them.

Start with the case the node was built for: a chain that reads something long and gives back something short. The template published as workflow 17933 turns research PDFs into seminar teaching packs, with HTTP Request pulling the files, an AI Agent driving the work, a Structured Output Parser shaping the result, and two chat models supplying the text. Swapping which model sits on that port changes nothing else in the workflow.

Second case, classification at volume. Text Classifier and Information Extractor both expect a model on the same port, and both behave better when the output stops drifting. That is what Sampling Temperature is for, and what Response Format set to json_object is for when the next node parses the answer as JSON rather than reading it as prose.

Third case, a question and answer chain over your own documents, where the model writes the final sentence and nothing else. Here Maximum Number of Tokens and Timeout matter more than creativity settings: a long context plus a slow answer is how an execution ends up cut short.

When to reach for something else. This node only talks to models on Qwen Cloud, so a workflow standardized on another provider is better served by OpenAI, Anthropic or Google Gemini. And if the prompt is a one-off with no tools and no memory, a Basic LLM Chain is a lighter root node than an AI Agent.

Known limits. There is no main input and no main output, so the node cannot be dropped between two steps of a workflow and never runs alone. It exposes one operation, generating a chat response, with no image, audio or embedding mode. The page covers version 1 of the node, the highest one in the catalog; an older workflow may show fewer options. Qwen Cloud bills its own API calls under its own terms, and n8n adds nothing on top. If the model side is new to you, the n8n training covers where sub-nodes fit.

Connect

How do you authenticate the node?

  1. 01

    Drop the node on the model port

    Open the root node you want to power, an AI Agent or a chain, and click the connector marked Model under it. n8n filters the list to sub-nodes that fit that port, so Qwen Cloud Chat Model appears there. The node opens with an empty credential field at the top, above the model selector. Nothing runs until that field is filled, because a sub-node has no main input of its own to execute from.

  2. 02

    Pick Gateway credits or your own key

    On n8n Cloud you can select Use Gateway credits in the node's credential field and run it without a Qwen Cloud account at all. Otherwise create a credential holding your own Qwen Cloud API key, from the Credentials menu or from the same field. A credential in n8n is a stored set of secrets, kept apart from the workflow so the key never sits in a node parameter.

  3. 03

    Choose the model and test the chain

    With the credential saved, the Model selector lists what your account can call, and the credential is reusable across every other workflow without being retyped. Then execute the root node, not this one. The answer, the token usage and any provider error surface in the root node's output panel, since the sub-node reports through whoever called it.

Connections

What can you plug this node into?

A port is the small connector under a root node that accepts one kind of sub-node. This one has a single output and no input, so the only question is which root node it feeds.

Output (what it plugs into)

  • Modelai_languageModel
01ModelThe node attaches to the model port of a root node and answers whenever that node needs text generated. Without a model on this port, a root node cannot run at all.

The node attaches to the model port of a root node and answers whenever that node needs text generated. Without a model on this port, a root node cannot run at all.

Key parameters

  • Required on the root nodes that expose the port: they have no fallback model of their own.
  • AI Agent and Basic LLM Chain are the usual starting points, the first when tools are involved, the second for a single prompt.
  • Information Extractor and Text Classifier accept the same port when the goal is structured output rather than conversation.
Use cases
feeding a Question and Answer Chain that answers from your own documents, or a Summarization Chain that condenses a long input before it reaches Slack.
Parameters

What does each parameter change?

The Qwen Cloud Chat Model node has 9 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 which Qwen Cloud model generates the completion. The dropdown reads from the credential, so the list reflects what your account can actually call rather than a hardcoded set.

Key parameters

  • Model: the only required choice in the node, and the one that decides the token ceiling every other setting has to live within.
Use cases
keeping a cheap fast model on a classifier that runs on every inbound message, and a stronger one on the chain that writes the final answer, both inside the same workflow.
02

Frequency Penalty

options.frequencyPenalty

What you see in n8n

Notes & use cases

Positive values penalize tokens based on how often they already appear in the text, which lowers the odds of the model repeating the same line word for word. A token is the unit a model reads and writes, roughly a word fragment.

Key parameters

  • Frequency Penalty: a number, default 0, meaning no penalty at all unless you raise it.
Use cases
a summarization chain that keeps restating the same sentence across bullet points, where a small positive value breaks the loop without touching temperature.
03

Maximum Number of Tokens

options.maxTokens

What you see in n8n

Notes & use cases

Caps how long the completion can get. The real ceiling depends on the model you selected, so this setting trims the answer rather than extending what the model supports.

Key parameters

  • Maximum Number of Tokens: a number, default -1, which leaves the model's own default limit in place.
Use cases
a chain writing a short reply that gets appended to a row in Google Sheets, where a capped answer keeps the cell readable and the call predictable.
04

Response Format

options.responseFormat

What you see in n8n

Notes & use cases

Decides the shape of what comes back, plain prose or a JSON object. Worth setting whenever the node after the chain parses the answer instead of reading it.

Key parameters

  • Text (text): a regular text response, the default.
  • JSON (json_object): JSON mode, which should guarantee the generated message is valid JSON.
Use cases
an extraction step whose output feeds a mapping into Gmail, where a stray sentence around the braces would break the field mapping.
05

Presence Penalty

options.presencePenalty

What you see in n8n

Notes & use cases

Close cousin of the frequency setting, with a different trigger: it penalizes tokens for having appeared at all, not for how often, which pushes the model toward new topics.

Key parameters

  • Presence Penalty: a number, default 0, so the behavior is neutral until you change it.
Use cases
a brainstorming chain asked for several angles on one subject, where the answer otherwise circles back to the first idea in every paragraph.
06

Sampling Temperature

options.temperature

What you see in n8n

Notes & use cases

Controls randomness. Lower it and the completions get less random; as it approaches zero the model turns deterministic and repetitive, which is exactly what you want when the output is parsed rather than read.

Key parameters

  • Sampling Temperature: a number, default 0.7, a middle setting that still varies from one run to the next.
Use cases
a Text Classifier that must return the same label for the same message every time, where anything but a near zero value makes the routing unstable.
07

Timeout

options.timeout

What you see in n8n

Notes & use cases

How long one request is allowed to run, in milliseconds, before the node gives up on it. Long prompts and long answers both push against this number.

Key parameters

  • Timeout: a number in milliseconds, default 360000, which is generous and rarely the first thing to tighten.
Use cases
a chain triggered by a form where a visitor is waiting on screen, so a shorter timeout returns an error you can handle instead of leaving the execution hanging.
08

Max Retries

options.maxRetries

What you see in n8n

Notes & use cases

Sets how many extra attempts the node makes when a request fails. Retries happen inside the node, so the root node sees one result and the workflow does not branch.

Key parameters

  • Max Retries: a number, default 2, meaning a failed call is already tried again twice before the error surfaces.
Use cases
a nightly batch where a single provider hiccup should not stop the run, paired with a timeout short enough that the retries still fit in the schedule.
09

Top P

options.topP

What you see in n8n

Notes & use cases

The other lever on diversity, through nucleus sampling: at 0.5, half of the likelihood weighted options are considered. The guidance is to alter this or the temperature, not both.

Key parameters

  • Top P: a number, default 1, which leaves the full set of options on the table.
Use cases
tightening a drafting chain that already runs at its default temperature, since moving one lever at a time is the only way to tell what changed the output.
Need help

Need help automating Qwen Cloud Chat Model with n8n?

A person reads every message.

FAQ

Common questions

01Is the n8n Qwen Cloud Chat Model node included with n8n?
Yes. It ships inside n8n's AI package, so there is nothing to install and nothing extra to pay on the n8n side, on n8n Cloud and on a self-hosted instance alike. Self-hosting runs through Docker or npm under the Community Edition and its Sustainable Use license, and a workflow built in one place behaves the same in the other. What sits outside that is the model itself: Qwen Cloud bills its own API calls under its own terms, and n8n adds nothing on top of them. On n8n Cloud you also have the option of Gateway credits in the node's credential field, which lets the node run without a Qwen Cloud account of your own.
02What do you need for the node to work?
Two things, and only two. A credential in the node's credential field, either Gateway credits on n8n Cloud or your own Qwen Cloud API key, and a root node to attach to. The node is a sub-node: it has no main input, so it never executes on its own and you always run the node it feeds instead. Once the credential is saved, the Model selector fills with what your account can call, and that same credential is reusable in every other workflow without retyping the key. No other field is required, since all 9 parameters beyond the model choice have working defaults.
03What are the limits of this node?
It does one thing: generate a chat-style response from a model available on Qwen Cloud. There is no image, audio or embedding mode, no main input and no main output, so it cannot be placed between two ordinary workflow steps. It only reaches Qwen Cloud models, so a different provider means a different sub-node on the same port. Version 1 is the highest in the catalog and the one described here, and an older workflow can show fewer options. The maximum token setting also depends on the model you selected rather than on the node.
04What do you have to plug in at minimum, and which sub-node should you start with?
At minimum, one root node and one model. Pick the root node for the job first: an AI Agent when the task needs tools, a Basic LLM Chain when a single prompt is enough, an Information Extractor or a Text Classifier when the output must be structured. Then attach this node to the Model port and fill the credential. If you are starting out, keep the defaults for everything except the model itself, run the root node, and read the output panel. Only then change one sampling setting at a time, because two changes at once tell you nothing about which one worked.
05n8n or Make for a chain like this?
It depends on where you want the workflow to live and how you want to pay for it. n8n runs on your own servers through Docker or npm, or on n8n Cloud, so prompts and documents can stay inside infrastructure you control. Make is hosted only, with no self-hosting option, and is billed per operation, which is predictable for steady volumes and less so for a chain that loops. The visual logic differs too: n8n's AI nodes work on the root node and sub-node model described here, where the model, the memory and the tools are separate boxes you attach. Try one chain in each and judge on your own volumes.
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