n8n pricing works on executions, not tasks. One execution is a single run of an entire workflow, no matter how many steps it contains — so a 40-step automation costs the same as a 2-step one. Cloud plans start at roughly €20/month (Starter, ~2,500 executions, billed annually) and scale to €50/month (Pro, 10,000 executions) and a Business tier around €667/month (40,000 executions, SSO). The self-hosted Community Edition is free and open source — you pay only for infrastructure and the engineering time to run it. Figures below are from n8n.io/pricing as of July 2026; verify live before you budget.
That execution-based model is the single most important thing to understand, because it inverts how most teams estimate cost. On Zapier or Make, complexity is expensive — every step burns quota. On n8n, complexity is nearly free and frequency is what you pay for. Below is how the plans actually break down, where the hidden costs sit (especially when you self-host), and how to decide which route fits an operations-heavy business.
How does n8n count an execution?
Per n8n's own definition, an execution is "a single run of your entire workflow. It doesn't matter how many steps are in the workflow or how much data it processes — it's still a single execution." This is the detail that trips people up — the top organic result for "n8n pricing" sits right next to a Reddit thread literally titled "N8N Pricing Confusion."
The practical consequences:
- Deep workflows are cheap. A single automation that pulls a record, enriches it across five APIs, branches, formats, and writes to three destinations is one execution. On per-task tools that same logic could be 10+ billable operations.
- High-frequency triggers are what add up. A workflow that fires on every inbound webhook, every new row, or every minute on a schedule is where your execution count climbs. Consolidate chatty triggers and batch where you can.
- Sub-workflows count too. If you call a sub-workflow, that's typically its own execution. Reusable "helper" workflows are great for maintainability but do register against your quota.
So the estimating question isn't "how complex is my automation?" — it's "how many times a month does it run?" Count your triggers and their frequency, not your nodes.
What do the n8n Cloud plans cost?
n8n Cloud is the managed, hosted option — no servers to run. The four published tiers (annual billing, in euros) look like this:
| Plan | Price (billed annually) | Executions / mo | Concurrency | Notable limits & features |
|---|---|---|---|---|
| Starter | ~€20/mo | ~2,500 | 5 concurrent | 1 shared project; unlimited users & workflows |
| Pro | ~€50/mo | 10,000 | 20 concurrent | 3 shared projects; unlimited users & workflows |
| Business | ~€667/mo | 40,000 | Higher | SSO / SAML / LDAP; 6 shared projects; self-host deployment option |
| Enterprise | Custom | Custom | 200+ concurrent | Unlimited projects; hosted or self-hosted; dedicated support |
Two honesty caveats worth stating plainly, because most pricing posts skip them:
- Month-to-month costs more than annual. The numbers above assume annual billing. Paying monthly runs higher, and the discount for committing annually is meaningful.
- Currency and repricing. n8n publishes in euros and has restructured this pricing more than once — third-party guides quoting the Business tier at "$800/month" aren't wrong, they're just a different currency and a different snapshot. Treat every number here as directional and confirm on the live pricing page before you commit budget.
The jump from Pro (€50) to Business (~€667) is steep and catches teams off guard. In practice you don't move to Business for the extra executions — you move for SSO/SAML and the governance features. If your blocker is "we need single sign-on and audit-grade access control," that's the line item you're actually buying.
Is self-hosted n8n really free?
The software is. The Community Edition is open source and free to run — that's genuinely the headline, and it's why n8n is popular with technical teams. But "free software" is not "free to operate," and this is where budgets go wrong. Self-hosting swaps a predictable subscription for an operating cost that you now own.
At ATI we frame the true cost of running any automation on our AI agent cost page as a three-layer system, and n8n self-hosting maps onto it cleanly:
The runtime layer (hosting) is the one everyone budgets for and the one that's genuinely cheap — a modest server runs a self-hosted instance fine at low volume. The operations and people layers are the ones that get underestimated. Someone has to own upgrades, watch for failed executions at 2 a.m., patch the box, and rotate credentials. When a critical workflow silently stops firing, the cost of that outage dwarfs any subscription you avoided.
That's the honest trade: self-hosting is cheapest in dollars and most expensive in attention. It wins when you have engineering capacity and a reason to keep data in your own environment. It loses when the team running it doesn't actually have room to operate it.
Self-hosted Community vs self-hosted Enterprise — what's the difference?
Confusingly, "self-hosted" isn't a single thing. There are two self-hosted editions:
- Community Edition — free, open source, the standard self-hosted n8n. Everything you need to build and run workflows.
- Self-Hosted Enterprise — a paid license that adds the governance layer on top of your own infrastructure: SSO/SAML/LDAP, external secret stores, log streaming, extended data retention, and dedicated SLA-backed support.
So the decision isn't binary "cloud vs self-host." It's a grid: managed vs self-run on one axis, community vs enterprise features on the other. Most operations-heavy SMBs start on Community self-hosted or Cloud Starter/Pro, and only reach for an Enterprise license when compliance or identity requirements force it.
Cloud or self-hosted — which should your business choose?
The deciding factors, in order of how often they actually settle it:
- Data residency / compliance. If regulated data can't leave your environment, self-hosting is the answer regardless of cost. This decides it more often than price does.
- Engineering capacity. No spare hands to run infrastructure? Cloud, full stop. The subscription is cheaper than a distracted engineer.
- Volume. At very high execution counts, self-hosting's flat infrastructure cost eventually beats per-tier cloud pricing — but only once you're paying for the ops discipline anyway.
- Governance needs. SSO, audit trails, and access control push you to Cloud Business or a self-hosted Enterprise license either way.
How does n8n pricing compare to Zapier and Make?
This is where the execution model pays off. Named honestly:
- Zapier charges per task — each step that runs is billable. Multi-step Zaps get expensive fast, and pricing scales with workflow complexity.
- Make charges per operation — similar dynamic, priced per module run, cheaper per unit than Zapier but still complexity-sensitive.
- n8n charges per execution — the whole workflow, once. Complexity is effectively free; you pay for frequency.
The takeaway isn't "n8n is always cheaper." It's that n8n gets relatively cheaper the more complex your automations are, and per-task tools stay competitive for simple, low-frequency zaps. If your workflows are deep — enrichment chains, multi-branch logic, heavy data transformation — n8n's model is structurally in your favor. If they're shallow and infrequent, the difference is smaller than the switching cost.
How should an operations team actually budget for n8n?
A practitioner's estimating sequence:
- Count trigger frequency, not steps. List each workflow and how many times a month it fires. Sum those — that's your execution estimate. Add headroom for sub-workflow calls.
- Map to a plan. Under ~2,500/mo, Starter. Under 10,000, Pro. Above that, or if you need SSO, Business or self-host.
- Price the layer you're avoiding. If you lean self-host to save the subscription, honestly cost the runtime, operations, and people layers above. Cheap software, real operating cost.
- Tie it to outcome, not tool spend. The number that matters isn't the n8n bill — it's hours recovered. Our AI ROI calculator models that directly: team size × weekly hours lost × 4.33 × cost per hour. If a €50/month Pro plan automates away 40 hours of manual work a month, the subscription is a rounding error.
That last point is the one implementation partners see constantly: teams optimize the €50 tool bill and ignore the €5,000 of manual labor the tool was supposed to eliminate. Budget the workflow's value first, then the license.
Where an implementation partner fits
Plenty of teams run n8n themselves and do it well — the platform is built for that. Where ATI tends to get pulled in is when the automation moves from "handy internal script" to production operations: when a failed execution means a customer doesn't get invoiced, when execution counts spike unexpectedly, or when the self-hosted instance no one has time to babysit becomes a single point of failure. We design, ship, and operate automation systems with the cost architecture, monitoring, and governance that keep them reliable at scale — the operations and people layers, handled. If you're weighing whether n8n is the right foundation or whether the ops burden makes managed delivery smarter, book a strategy call and we'll map it against your actual workflows.
Frequently asked questions
How much does n8n cost per month?
n8n Cloud starts around €20/month for the Starter plan (~2,500 executions, billed annually), €50/month for Pro (10,000 executions), and roughly €667/month for Business (40,000 executions, with SSO). Enterprise is custom-priced. The self-hosted Community Edition is free — you pay only for infrastructure and operations. Confirm current figures at n8n.io/pricing before budgeting, as they're published in euros and change periodically.
What counts as an execution in n8n?
One execution is a single run of an entire workflow, regardless of how many steps it contains or how much data it processes. A 40-step workflow and a 2-step workflow each count as one execution per run. This is different from Zapier (per task) and Make (per operation), which bill for each step.
Is self-hosted n8n free?
The Community Edition software is free and open source. Operating it is not: you pay for hosting/infrastructure, plus the engineering time to maintain, monitor, secure, and upgrade it. For low-volume use with available engineering capacity it's very cost-effective; without that capacity, Cloud is usually cheaper in total.
Is n8n cheaper than Zapier?
It depends on your workflows. Because n8n charges per full execution rather than per step, it gets relatively cheaper as automations grow more complex. For deep, multi-step workflows n8n's model is structurally cheaper; for simple, infrequent automations the gap narrows and switching cost may outweigh savings.
Which n8n plan do I need for SSO?
SSO/SAML/LDAP is available on the Cloud Business plan and on the self-hosted Enterprise license. If single sign-on and audit-grade access control are your requirement, that's the tier you're buying — not the extra execution capacity.
How do I estimate my n8n execution usage?
Count how often each workflow runs per month based on its trigger frequency — not the number of steps. Sum those runs, add headroom for any sub-workflow calls (which count separately), and match the total to a plan tier. Frequency drives cost far more than complexity does.