SaaS Radar

What Does n8n Pricing Look Like in 2026? Cloud vs Self-Hosted

Published 2026-10-01 · Researched and price-checked on official sources

n8n Cloud costs €20/mo billed annually on Starter for 2,500 workflow executions, €50/mo on Pro for 10,000 and €667/mo on Business for 40,000, according to n8n pricing published on 1 October 2026. Self-hosting the Community Edition replaces that subscription with a server bill.

Key facts

What matters

n8n Cloud runs €20/mo (Starter, 2,500 executions), €50/mo (Pro, 10,000) and €667/mo (Business, 40,000) when billed annually, per n8n's pricing page on 1 October 2026.
A workflow that polls every five minutes executes 8,600 to 8,900 times a month, more than triple the Starter allowance, and n8n's own FAQ gives that figure.
The Community Edition is free to run on your own server, but the Sustainable Use License covers internal business use only: you may not sell n8n hosting to clients without a commercial licence.
n8n's deployment docs ask for at least 4 GB RAM and 2 vCPU, which rules out the cheapest 1 vCPU VPS tiers even when they advertise 4 GB.

What n8n is, and what its pricing actually covers

n8n is a source-available workflow automation platform that mixes code steps, integrations and AI agent nodes in one editor. It is sold two ways, and that split is the whole pricing story: a hosted subscription billed by workflow executions, or free software you install on a server you rent.

Because both paths are legitimate, comparison searches tend to mix them up. The cloud tiers compete with Zapier and Make on monthly execution allowances. The self-hosted path competes with a VPS invoice plus your own maintenance time. The numbers below come from n8n's own pricing page and deploy docs, checked on 1 October 2026.

n8n pricing in 2026: the four cloud plans, verified

Starter, €20/mo billed annually: 2,500 workflow executions, one shared project, 5 concurrent executions, AI agent and MCP nodes, and 1,600 Assistant credits a month. It is the tier to test an idea on, not to run a business on.

Pro, €50/mo billed annually: 10,000 executions, three shared projects, up to 50 concurrent executions, 7 days of insights, and up to 9,600 Assistant credits. n8n positions this for solo builders and small teams in production.

Business, €667/mo billed annually: 40,000 executions, six shared projects, SSO/SAML/LDAP, 30 days of insights, queue mode for scaling, and Git-based version control. n8n describes it for companies under 100 employees. Businesses with fewer than 20 staff can ask about n8n's startup program, which the pricing page lists at 50% off Business.

Enterprise, contact sales: custom execution volumes, 200+ concurrent executions, 365 days of insights, log streaming, external secret store integration and a support SLA.

Note the billing toggle. The figures above are the annual-billing rates n8n displays, and switching to monthly billing raises the effective monthly rate. The page states the annual option saves 17%.

How n8n counts an execution, and why your schedule decides the bill

n8n bills one execution for a complete workflow run, no matter how many steps it contains. That is genuinely friendlier than per-task billing: a 30-node workflow is still one execution. The catch is that a single chatty schedule can consume an entire tier.

n8n's own FAQ does the arithmetic for you: a workflow scheduled daily runs 30 or 31 times a month, while one that runs every five minutes lands at roughly 8,600 to 8,900 executions monthly. Put that polling workflow on Starter and you have blown through 2,500 executions three times over before the month ends.

A daily digest workflow is nearly free at any tier.
Webhook-triggered workflows consume one execution per event, so traffic spikes become the bill.
An always-on agent that checks a queue every few minutes is the pattern that pushes buyers from Starter straight to Business.

Assistant credits are separate from executions and less forgiving: the FAQ states unused credits do not roll over, and n8n does not yet sell top-up credits. Running out means upgrading the plan, not buying 2,000 more.

Self-hosted n8n: free software, a real server bill

The Community Edition is the same product installed on your own machine, and n8n's pricing page points to GitHub for it. The licence matters here. n8n ships under the Sustainable Use License, which grants use for internal business purposes or personal work and allows redistribution only for free, non-commercial use. An agency cannot host n8n for paying clients on the free licence, which is exactly the gap the Business and Enterprise plans fill.

Hardware is the other half. n8n's deployment guide for hosting on a cloud provider tells you to provision at least 4 GB RAM and 2 vCPU for a Docker install with the code sandbox running. That spec matters more than it looks: budget VPS pages advertise a 4 GB plan with a single vCPU, which misses the CPU half of the requirement.

For reference, the entry points verified on 1 October 2026:

Hostinger KVM 1: $6.49/mo intro, renewing at $11.99/mo on a two-year term for 1 vCPU, 4 GB RAM, 50 GB NVMe and 4 TB bandwidth. Enough RAM, one core short of n8n's guidance.
Hostinger KVM 2: $8.99/mo intro, renewing at $14.99/mo, with 2 vCPU, 8 GB RAM, 100 GB NVMe and 8 TB. This is the cheapest plan that meets the documented 2 vCPU bar.
RackNerd: KVM VPS plans are advertised from $2.24/mo, the cheapest starting point on this list, with annual prepay and thinner support.

Add your model API keys on top of the server. Self-hosted n8n does not include Assistant credits; n8n's FAQ says self-hosted Assistant runs in preview with your own key, which removes the credit limit and moves the AI cost onto your provider bill.

Trials, and how to size your execution count before you pay

Both routes start free, with different strings attached. The n8n Cloud trial needs no credit card, while the Business trial does require one and runs 14 days with all Business features unlocked. Either way the trial includes 800 Assistant credits that expire when it ends.

Sizing the right tier is the step most buyers skip, and n8n does publish the method. If you already run n8n, read your execution logs or the Insights Dashboard; Business customers also get a weekly email with production execution counts across every instance holding their licence key. If you are starting fresh, count your intended schedules the way the FAQ does: daily jobs cost about 30 executions a month, five-minute polling costs about 8,700.

That single multiplication usually decides the plan. A team migrating 20 daily automations needs under 1,000 executions a month and belongs on Starter. A team migrating one queue-polling agent needs a plan 3.5 times larger than that, which is where the jump from €20 to €667 starts to look rational.

The costs n8n's pricing page does not put in the headline

Three terms in the fine print are worth knowing before you commit to an annual plan.

Overage on Business is priced per bucket. n8n's FAQ states that exceeding your quota does not stop workflows, but overages on the Business plan are billed at €4,000 per additional 300,000 executions and invoiced automatically 45 days after they occur.
No pro-rata refunds. Annual Business subscriptions run to the end of the annual term when cancelled, per the same FAQ.
Data residency is Frankfurt. n8n states that hosted plans store data in the EU, on servers in Frankfurt, Germany. Self-hosted deployments live wherever you put them, which is the usual reason EU buyers self-host at all.

Enterprise-only items are also worth scanning for: log streaming, external secret stores and 365-day insight retention are not available on the tiers below it, so a compliance requirement pushes the budget past €667/mo without a listed price.

Which n8n plan fits your workload

Use one daily or webhook-driven workflow with modest traffic: Starter at €20/mo billed annually is the correct entry point, and it costs less than the time you would spend maintaining a server.

Run a handful of production automations with a team: Pro at €50/mo, mainly for the 10,000 executions, three projects and 7-day insight window.

Poll frequently, run agents around the clock, or need SSO and queue mode: Business is the tier n8n built for that load, and the startup discount for companies under 20 employees is worth asking about before you pay €667/mo.

Want the lowest cash cost and do not mind operations: self-hosting on a 2 vCPU / 8 GB VPS starts near $8.99/mo intro at Hostinger plus your model costs, but you own Docker, PostgreSQL, backups and upgrades. If you plan to host n8n for clients, budget for a commercial licence instead of the Community Edition, because the Sustainable Use License does not permit that.

At a glance

n8n plan	Price, billed annually	Executions per month	Best fit
Starter	€20/mo	2,500	First workflows and one small agent
Pro	€50/mo	10,000	Solo builders and small teams in production
Business	€667/mo	40,000	Under 100 staff, needs SSO/SAML and queue mode
Enterprise	Contact sales	Custom	Compliance, log streaming, 365-day insights
Community Edition (self-hosted)	Free licence plus server cost	Not metered by n8n	Teams that will run Docker, PostgreSQL and upgrades

n8n Cloud plans as listed on n8n.io/pricing, checked 1 October 2026. Prices are the annual-billing rates.

FAQ

Is n8n free?

The Community Edition is free to self-host from GitHub, licensed under the Sustainable Use License for internal business or personal use. n8n Cloud is not free: Starter is €20/mo billed annually with 2,500 executions, per n8n's pricing page on 1 October 2026. Self-hosting still costs you the server and any model API keys.

What happens if I exceed my n8n execution quota?

Your workflows keep running. n8n's FAQ says overages may be charged, and on the Business plan they are billed at €4,000 for each additional 300,000 executions, invoiced 45 days after the overage. Upgrading to the next tier is usually the cheaper fix.

Is n8n Cloud or self-hosted cheaper?

Self-hosting is cheaper in cash once you pass Starter's 2,500 executions: a 2 vCPU, 8 GB VPS starts near $8.99/mo intro at Hostinger against €667/mo for Business, and you pay your own model costs. Cloud is cheaper in time, and n8n stores hosted plan data in Frankfurt, Germany, so EU data residency is already handled.