Contents
At a glance
- Unit
- One complete workflow run, trigger to finish
- Node count
- Irrelevant - a 2-node and a 200-node workflow both count as one
- Data volume
- Irrelevant to the execution count
- What drives cost
- Trigger frequency and schedule granularity
- Cloud Starter
- EUR 20/month for 2,500 executions (checked August 2026)
- Cloud Pro
- EUR 50/month for 10,000 executions (checked August 2026)
- Self-hosted
- No execution meter - you pay infrastructure and maintenance instead
What one execution is
An n8n execution is a single run of an entire workflow. n8n's wording is unambiguous: it does not matter how many steps are in the workflow or how much data it processes, it is still a single execution. A two-node workflow and a two-hundred-node agent both count as one.
This is the opposite design to per-action billing. On n8n the expensive thing is how often the workflow fires. Node count, loop iterations, records processed and run duration are all free at the execution meter. n8n markets this directly, contrasting it with tools that charge per step or per user.
What makes the count go up
Three things. First, trigger frequency: a webhook that fires on every record change will out-consume a nightly batch by orders of magnitude for identical work. Second, schedule granularity: a cron set to every minute is 43,200 executions a month before it has done anything useful. Third, workflow splitting - if you break one workflow into a parent and a called sub-workflow, you have created a second run, and how it is metered depends on your plan and version, so check your own execution log rather than assuming.
The design move that follows is batching. If you can make one run handle a hundred records instead of firing a hundred times, you have cut your bill by 99% and changed nothing else. On Zapier the same change would save you almost nothing, because there you pay per action regardless.
Cloud plans and limits
As published on n8n's pricing page in August 2026, n8n Cloud lists Starter at EUR 20 a month for 2,500 executions and Pro at EUR 50 a month for 10,000 executions, with Business at EUR 667 a month for 40,000 executions and Enterprise on custom terms. Prices are quoted in euros on that page and the annual figures are shown; check the page for your currency and billing period before budgeting.
Self-hosting removes the execution meter entirely and replaces it with infrastructure and your own time. That is a real trade and not always the cheaper one - see the self-hosting entry and our self-hosting cost calculator for the arithmetic.
What you should know
Two costs sit outside the execution meter and catch people out. Model tokens are billed by the model provider, not by n8n, so an AI agent node that costs one execution can still cost real money in tokens. And execution data retention is a separate dimension: keeping full execution data for debugging has storage implications on cloud plans and is worth checking against your own plan.
Failed executions still count as executions on most metered systems, so a workflow retrying in a loop against a down API can burn allowance while accomplishing nothing. Pair retry and backoff with a cap, and set error handling that stops rather than loops.
n8n execution FAQ
What counts as one execution in n8n?
One full run of a workflow. n8n states that it does not matter how many steps the workflow has or how much data it processes - it is still a single execution. This is what makes large, complex workflows cheap on n8n relative to per-step platforms.
Does a loop in n8n cost extra executions?
No. Iterating over 500 items inside one workflow run is still one execution. This is the single biggest structural difference from per-action or per-operation pricing, where the same loop would be 500 billed units per step inside it.
How many executions do n8n Cloud plans include?
As listed on n8n's pricing page in August 2026: Starter EUR 20 a month for 2,500 executions, Pro EUR 50 a month for 10,000, and Business EUR 667 a month for 40,000. Enterprise is custom. Verify the current figures and your billing currency on the page itself.
Do failed n8n executions count?
Treat them as counting. A workflow that starts and errors has still run, and a retry loop against a failing API can consume allowance without producing any result. Cap retries and use error handling that stops the run rather than reattempting indefinitely.
Is self-hosted n8n free?
There is no execution meter on the self-hosted community edition, but it is not free of cost. You take on hosting, upgrades, backups, queue mode configuration at volume, and the time to keep it running. Whether that is cheaper than a cloud plan depends on your volume and your hourly rate.
Do AI nodes cost more executions?
No, but they cost money elsewhere. An AI agent node is part of the same single execution, while the tokens it consumes are billed separately by whichever model provider you pointed it at. Budget the platform meter and the model bill as two different lines.
Primary source
n8n - pricing and execution definition
Checked August 2026. Anything priced or rate-limited moves, so verify against the source before you budget.
Want this built rather than explained?
Knowing what n8n execution means is one thing; having it running in your business is another. We scope one workflow at a time, quote a fixed price in writing after a free consult, build it, and hand it over with documentation you own.
