At a glance
- Limit each callOpen the model node's options (for example OpenAI Chat Model) and set Timeout to what a normal answer takes, plus a margin, such as 60 seconds.
- Limit the whole runIn the workflow's Settings, turn on Timeout Workflow and set a limit.
- Hear about itSet an error workflow in the workflow's Settings that sends the workflow name and execution link to email or Slack.
Limit each call
- Open the model node's options (for example OpenAI Chat Model) and set Timeout to what a normal answer takes, plus a margin, such as 60 seconds.
- Set Max Retries low, one or two. Retrying a provider that is down only makes the wait longer.
Limit the whole run
- In the workflow's Settings, turn on Timeout Workflow and set a limit. A run that passes it is stopped and marked as failed.
- On self-hosted n8n, EXECUTIONS_TIMEOUT sets a default for every workflow, and EXECUTIONS_TIMEOUT_MAX caps what anyone can set.
Hear about it
- Set an error workflow in the workflow's Settings that sends the workflow name and execution link to email or Slack.
- For important steps, use the node's On Error setting to continue on a second branch, for example a cheaper model or a queue to try later, instead of stopping.
- Review long runs weekly in Executions, sorted by duration.
Updated 2026-11-01. General guidance, not advice for your situation; we scope each project in writing before any work starts.
Questions people ask first
What timeout should I pick?
Look at a week of runs and take the slowest normal one, then add half again.
Will a timeout lose the item?
The run is marked as failed. With an error workflow you get the execution link and can retry it.
Does this apply to the Agent node?
Yes. Set the timeout on the chat model connected to the agent, and the workflow timeout on top.
Tell us what you need built.
Tell us the process and the tools involved, or call. We reply with questions or a written scope.
hello@pushbuild.com