n8n · pricing explained · last verified 4 September 2026

n8n pricing explained

n8n pricing is based on workflow executions — not on steps, tasks or users. As of 4 September 2026, n8n Cloud has two self-serve plans: Starter at 20€ a month with 2,500 executions and Pro at 50€ a month with 10,000, both billed annually. The Business plan is 667€ a month for 40,000 executions and is a self-hosted license; Enterprise is priced on request. Every plan includes unlimited users, unlimited workflows and every integration. Self-hosting the Community edition is free — no license key, and n8n’s pricing page lists execution quotas only for the paid plans.

One execution is one run of an entire workflow, however many nodes it has. Below: every plan in one table, what a single run actually costs on each, how many executions our own templates burn at different trigger cadences, and the point at which self-hosting becomes the cheaper option.

n8n plans at a glance (4 September 2026)

PlanPriceExecutions / monthHostingConcurrent executionsProjectsAI creditsAlso included
Starter 20€/mo, billed annually 2.5K Hosted by n8n 5 1 shared project 2,300 AI credits/month Unlimited users · Forum support
Pro 50€/mo, billed annually 10K Hosted by n8n 20 3 shared projects Up to 13,700 AI credits/month 7 days of insights · Admin roles · Global variables · Workflow history · Execution search
Business 667€/mo, billed annually 40K Self-hosted not listed on the plan card 6 shared projects AI Assistant (preview) SSO, SAML and LDAP · 30 days of insights · Different environments · Scaling options · Version control using Git · Forum support
Enterprise Contact Sales Custom Hosted by n8n or Self-hosted 200+ Unlimited shared projects N/A 365 days of insights · External secret store integration · Log streaming · Extended data retention · Dedicated support with SLA · Invoice billing

Copied from n8n.io/pricing on 4 September 2026, which displayed prices in EUR with the “Annually (Save 17%)” toggle selected. n8n also invoices in USD — its wire-transfer contract forms come in € and $ versions. Prices and quotas change; the source of truth is n8n’s page.

Execution limits by plan

LimitStarterProEnterprise
Max active workflowsUnlimitedUnlimitedUnlimited
Concurrent executions520200+
Max saved executions2.5k25k50k
Saved-execution storage2.5 GB25 GB50 GB
Max execution log retention7 days30 daysUnlimited
Workflow history – max days15365+

From the “What’s included?” table on the same page, which compares Starter, Pro and Enterprise only. Reaching a saved-execution or retention limit does not stop workflows — n8n removes the oldest executions.

What counts as an execution

n8n’s own definition, from the pricing FAQ: 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.

That is the whole difference from Zapier’s tasks and Make’s operations, which meter every step. In n8n a workflow that fetches an API, loops over 500 items and posts 500 Slack messages is one execution. The counter is the trigger: every time the trigger fires, one execution is used — whether or not the run finds anything to do.

n8n’s guidance for estimating your number (same FAQ): a scheduled workflow on a daily schedule is 30 or 31 executions a month; one that runs every 5 minutes is about 8,600–8,900; for a webhook, the number of times the triggering event occurs per day equals your daily executions; for a chatbot, conversations per week × messages per conversation.

What one workflow run costs

Divide the plan price by the executions it includes. These are our calculations from n8n’s numbers, not figures n8n publishes:

PlanArithmeticCost of one execution
Starter20€ ÷ 2,500 executions0.8 cents per execution
Pro50€ ÷ 10,000 executions0.5 cents per execution
Business667€ ÷ 40,000 executions≈1.7 cents per execution (you are also buying SSO, environments, Git and a self-hosted license)
Business overage4,000€ ÷ 300,000 extra executions≈1.3 cents per execution

The plan is flat, so the useful question is not “what does one run cost” but “what share of my allowance does this workflow eat”. Here is that share for the templates in our n8n template library, all of which were built and run on a real n8n instance. Scheduled rows use a 30-day month; webhook rows assume the monthly event volume shown.

TemplateTrigger and cadenceExecutions / monthShare of Starter (2,500)Share of Pro (10,000)
Read an RSS feed Schedule Trigger, every 15 minutes 2,880 115% — over the plan 29%
Read an RSS feed Schedule Trigger, hourly 720 29% 7%
Fetch and extract API data Schedule Trigger, once a day 30–31 1% 0.3%
Webhook form validation API Webhook — assume 300 submissions 300 12% 3%
Webhook JSON API endpoint Webhook — assume 1,000 requests 1,000 40% 10%
Webhook event router Webhook — assume 5,000 events 5,000 200% — over the plan 50%
  • Polling is the budget killerA Schedule Trigger burns an execution every tick even when the feed has nothing new. One RSS poll every 15 minutes (2,880 runs) exceeds the whole Starter plan on its own; n8n’s own example of an every-5-minutes schedule (8,600–8,900 runs) is most of a Pro plan for a single workflow. Poll hourly or daily unless minutes matter.
  • Webhooks only pay for real eventsThe webhook JSON API and the event router cost one execution per incoming request, so a quiet endpoint costs nothing. Prefer a webhook over polling wherever the source can push.
  • Node count is irrelevantAdding a Code node, a Set node and three more app nodes to a workflow changes nothing on the bill. Optimise the trigger cadence, not the node count.
  • Execution history is a separate limitStarter keeps 2.5k saved executions for 7 days, Pro 25k for 30 days. A high-frequency workflow can push older runs out of the log long before it touches the monthly quota.

Which plan — by monthly executions

Your volumePickWhy
Up to 2,500 executions a monthStarter — 20€/mo (annual)The cheapest managed option. 5 concurrent executions and a 7-day execution log.
2,500 to 10,000Pro — 50€/mo (annual)20 concurrent executions, a 30-day log, admin roles, global variables, workflow history.
More than 10,000Self-hosted Community edition (free) — or Business at 667€/mo if you need SSO, environments or projectsOn n8n Cloud the plan above Pro is Enterprise (contact sales). Business is a self-hosted license, so at this volume you are running a server either way.

Cloud vs self-hosted: the real total cost

The n8n docs are explicit that every self-hosted installation runs the same product: without a license key it is the free Community edition, which “includes almost the complete feature set”. What it leaves out is a fixed list — custom variables, environments, external secrets, external storage for binary data, log streaming, multi-main mode, projects, SSO (SAML, LDAP), workflow and credential sharing, and version control using Git. Registering your email unlocks folders, debug-in-editor and custom execution data, still free.

What you take on instead is also stated by n8n: setting up and configuring servers and containers, managing application resources and scaling, securing servers and applications, and configuring n8n — with the warning that it recommends self-hosting for expert users and that mistakes can lead to data loss, security issues and downtime. Your total cost is therefore the server plus the hours, and the hours are the larger number for most small teams.

  • Self-host whenyou are above 10,000 executions a month, need data to stay on your own infrastructure, want custom nodes or bash scripts (the pricing table marks both “available in self-hosted”), or already run Docker for other things. Our self-hosted n8n guide has the exact install commands from the docs.
  • Stay on Cloud whenyour volume fits Starter or Pro, nobody on the team wants to be on call for a server, or you need the Cloud-only conveniences n8n handles (hosting, updates, scaling). At 20–50€ a month, Cloud costs less than a few hours of maintenance.
  • Business is not “cheaper Cloud”It is a 667€/month license for a server you run, bought for SSO, environments, projects and Git — not for executions.

The fine print worth knowing

  • Annual billing is what the headline prices meanThe plan cards read “billed annually”, and the page labels annual billing “Save 17%”. Use the Monthly toggle on n8n’s page for the month-to-month rate.
  • The free trial is a trial, not a free tier1,000 executions, 5 concurrent executions, a 180-second execution timeout and 1 team project, with Pro-level features. No credit card for the Cloud trial; the Business trial (14 days) needs one.
  • AI credits do not roll over and cannot be bought2,300 a month on Starter, 5,700 or 13,700 on Pro depending on plan size. Unused credits expire monthly; to get more you upgrade. Self-hosted n8n uses your own key, so no credit limit applies.
  • Business plan means forum supportn8n describes Business as a self-serve option; dedicated support with SLAs is Enterprise only.
  • Hitting a quota does not stop your workflowsFor Business and Enterprise, workflows keep running and overage charges may apply — on Business, 4,000 EUR for each extra bucket of 300,000 executions, invoiced 45 days after they occur if there is no upgrade.
  • Annual commitments end at term, with no pro-rata refundsMonthly commitments can be cancelled at any time and end at the close of the billing cycle.
  • A paid self-hosted license phones homeBusiness and Enterprise license keys must ping n8n’s license server daily and report the number of production executions. Telemetry is on by default and can be disabled.
  • Where your data livesHosted plans store data within the EU, on servers in Frankfurt, Germany. Self-hosted: wherever you run it.
  • Start-up discountUnder 20 employees? n8n’s Start-up Plan offers 50% off Business.

Sources and last verified

Every number on this page was read from these pages on 4 September 2026. If n8n’s page and this one disagree, n8n’s page wins — tell us at support@flowtemplates.app and we will re-verify.

Comparing platforms, not plans? See n8n vs Zapier vs Make. Ready to try a workflow for free: the RSS feed reader and API fetch templates need no credentials at all.

Frequently asked questions

How much does n8n cost?

As of 4 September 2026, n8n Cloud Starter is 20€ a month with 2,500 workflow executions and Pro is 50€ a month with 10,000, both billed annually. The self-hosted Business plan is 667€ a month with 40,000 executions, and Enterprise is priced on request. Self-hosting the Community edition is free.

What counts as one execution in n8n?

n8n’s pricing page defines it as “a single run of your entire workflow. It doesn’t matter how many steps are in the workflow or how much data it processes.” A 3-node webhook and a 30-node pipeline each cost one execution per run.

Is n8n free?

The self-hosted Community edition is free, needs no license key and, per the n8n docs, includes almost the complete feature set. n8n Cloud has no free tier — only a trial with 1,000 executions, 5 concurrent executions and a 180-second execution timeout, no credit card required.

What happens when I use up my monthly executions?

For Business and Enterprise, n8n says workflows continue running without interruption and overage charges may apply — on Business, 4,000 EUR per extra bucket of 300,000 executions. The page does not publish an overage rate for Starter or Pro, so on those plans budget for the next tier.

Is self-hosting n8n cheaper than n8n Cloud?

Above 10,000 executions a month, staying on n8n’s own hosting means Enterprise (contact sales) and the self-serve Business plan is a 667€ self-hosted license; the Community edition costs only your server and your time. Below 10,000 executions, Starter or Pro at 20–50€ a month is usually cheaper than a server you maintain yourself, given the server, security and scaling skills n8n says self-hosting requires.

Does n8n charge per user or per step?

No. Every plan includes unlimited users and workflows and every integration; you pay for workflow executions. The one metered extra is AI Assistant credits — a monthly allowance (2,300 on Starter, up to 13,700 on Pro) that does not roll over and cannot be topped up.