LLM Basic Chain output format

In last versions the output format LLM Basic Chain was changed. Why?

Yesterday I noticed that the response format has changed. Now it returns json format
{response:
{text: “response llm”} }

In previous versions the output was of a different format, without “response”
{text: “response llm”}

Now everyone who has this node running in production will get NULL when processing the results. ($json.text is null). How did this happen?

Describe the problem/error/question

What is the error message (if any)?

Please share your workflow

(Select the nodes on your canvas and use the keyboard shortcuts CMD+C/CTRL+C and CMD+V/CTRL+V to copy and paste the workflow.)

Share the output returned by the last node

Information on your n8n setup

  • **n8n version: 1.83.2
  • Database (default: SQLite):
  • n8n EXECUTIONS_PROCESS setting (default: own, main):
  • **Running n8n: Docker
  • **Operating system: Ubuntu 22.04

Hey @korovaevda

As far as I know, nodes have versions.

So it’s not a problem at all.
Nodes running in production would only get the new output format if they created another node from scratch.

.
:point_right: If my reply answers your question, please remember to mark it as a solution.

Thanks for the answer!
I want to clarify. Do I understand correctly that if I’ve a workflow running in production, and update n8n, the data format in the nodes of this workflow should not change?
If I create a new one in the updated version of n8n, will it already be a new version and will it be possible to output a different data format?

Yes, that is correct.

Updating n8n shouldn’t change the version of the nodes that already exist.

What can happen, though, is some other bug related to changes of the new n8n version.

Updating n8n always opens the door for a possible bug to happen in an old workflow.

But that possibility is mitigated by keeping the node version intact.

Hey @solomon

I am not sure if there is something else going on here. This also started happening to me this morning after updating from 1.84.1 to 1.85.0

This is a daily workflow I have that has been running untouched for a while. Following screenshots show the difference between yesterday and today. But this is happening in all my Basic LLM chain nodes across all my workflows.

This is from the execution logs. Notice both have the same node version.

What its weird to me is that after downgrading back to 1.84 I am experiencing the same behavior.

I have look in github but there is no issue so far related to this, so I decided to start here.

1 Like

Hey @willy

So this is the case of something that changed beyond the node itself.

Maybe they updated the API request that is made, maybe they changed how every AI tool works. I don’t know.

But since this is the case, updating your workflows will be the only solution if you want to upgrade the n8n version.

That’s why it’s good to have a testing environment, so you can test newer n8n versions with your old workflows, for example.

:point_right: If my reply answers your question, please remember to mark it as a solution.

I had the same issue — I updated n8n last night, and 43 of my workflows broke due to the change in the output (response).

It seems this bug was fixed in version 1.85.3

1 Like

1.84.2 should already have this fix.

Updating the n8n version to the latest should fix, guys.

@Manuel_Llop @willy

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.