n8n: Cloud vs Self-Hosted, Pricing & Alternatives
Workflow automation platform for technical teams and AI agents
n8n is a workflow automation platform that supports app integrations, custom logic, self-hosting options, and AI agent workflows.
Visit n8nDirect answer
Choose n8n when a technical owner needs visual workflows with custom code, API calls, and a choice of hosting. Compare Make when the priority is managed app-to-app automation without operating an instance. Neither choice removes the need to handle failed runs and check AI output.
What changes between Cloud and self-hosting?
n8n combines a visual workflow canvas with code and integrations. Its official repository describes AI workflows with tool use and human approvals, plus a choice of Cloud or self-hosted deployment. The project is source-available under its published licenses; do not assume that every feature or commercial use is covered by Community access.
Choose Cloud
Start here if you want to build the workflow without also maintaining its server. Check the paid plan's execution allowance and retained history.
Consider self-hosting
Useful when your team can own installation, updates, backups, availability, and recovery. Include that work in the cost comparison.
Check the data path
Hosting n8n yourself does not keep data local if a workflow sends it to an external model or application. Inspect each connected destination.
Pricing: executions, hosting, and AI bills
The official pricing page checked September 7, 2026 lists Cloud Starter at €20/month billed annually for 2,500 workflow executions, and Pro at €50/month billed annually for 10,000. These are annual-billing equivalents in euros, not monthly USD quotes.
The main billing unit is a workflow execution: a complete run, rather than each individual step. Community Edition is available for self-hosting; infrastructure and external API costs remain separate. A Cloud trial is different from ongoing free managed hosting.
Example to budget: a workflow scheduled once a day has roughly 30 scheduled runs in a 30-day month before additional triggers or re-runs. Record external model charges separately. Do not equate an execution allowance with unlimited AI tokens.
Decision table
| Your requirement | What to check |
|---|---|
| Custom application logic | Build one representative API or code step and test missing fields, unexpected responses, and retries. |
| Self-hosted instance | Name the person responsible for updates, credentials, backups, and restoring a failed instance. |
| Managed visual automation | Compare Make on the same workflow and its measured credit consumption. |
| Lead enrichment | Compare Clay when the bottleneck is company/contact data rather than general orchestration. |
A useful first workflow
Suggested pilot: receive a sample support request, check required fields, create a draft summary, and route it to a human reviewer. Use a test destination so repeated runs cannot send duplicate customer messages.
- Try a valid request, missing data, and a deliberately failed integration.
- Check what the next run does with an item already processed.
- Compare the draft with the source request before allowing external delivery.
- Measure successful outcomes, recovery effort, and total cost together.
Skip self-hosting for now if no one will maintain it. A small Cloud pilot or an existing integration can answer the workflow question first.
FAQ
Is n8n free?
A Community Edition is available for self-hosting. That is different from free managed Cloud hosting; infrastructure and connected services can still cost money.
Does n8n charge for every workflow step?
Its paid pricing is primarily based on complete workflow executions. Check the selected allowance and any separately billed AI or connected services.
When should I compare Make instead?
Compare Make when you prefer managed visual app connections. Evaluate both with the same workflow, including failed runs and total usage cost.
Review basis and sources
Official sources checked September 7, 2026. This is an editorial assessment, not a hands-on benchmark. The suggested pilots above have not been run for this review; no reliability, cost-saving, or conversion result is claimed.
For a factual correction or relevant placement request, use our submission form or placement information. Payment does not guarantee a recommendation.