n8n Cloud
A workflow built in n8n Cloud, n8n's hosted service: nodes run in order from a trigger, and AI nodes call the model companies whose credentials you add. This describes a kind of platform, not anyone's instance; a self-hosted n8n is a server, with its own disk. n8n, which runs the instance, sees everything the workflow handles, and the fit has no field for that. Checked against n8n's documentation on 22 October 2026.
- Off for up to
- no time: it is always on
A hosted service that is always on. A Schedule Trigger fires only once the workflow is saved and published. https://docs.n8n.io/integrations/builtin/core-nodes/n8n-nodes-base.scheduletrigger/
- Runs steps as
- code
The workflow's nodes run in the order drawn, the same way every time. An AI Agent node is a step an AI decides, calling a chat model, so it counts under calls, not agents. https://docs.n8n.io/integrations/builtin/cluster-nodes/sub-nodes/
- AI from
- none
No agent runs over the whole instance: each AI node sees only what the workflow passes it. https://docs.n8n.io/integrations/builtin/cluster-nodes/sub-nodes/
- Can call
- any company
Chat model nodes reach OpenAI, Anthropic, Google Gemini and Vertex, Mistral, Amazon Bedrock and many more, and the HTTP Request node can call any API. https://docs.n8n.io/integrations/builtin/cluster-nodes/sub-nodes/ https://docs.n8n.io/integrations/builtin/core-nodes/n8n-nodes-base.httprequest/
- Holds
- nothing
n8n Cloud keeps no folders: its file system is limited to /home/node and files written there are not kept between executions, and the Execute Command node isn't available. https://docs.n8n.io/integrations/builtin/core-nodes/n8n-nodes-base.readwritefile/ https://docs.n8n.io/integrations/builtin/core-nodes/n8n-nodes-base.executecommand/