n8n pricing in 2026 starts at €20 a month (billed annually) for Cloud Starter with 2,500 executions, €50 for Pro with 10,000, and €667 for the self-hosted Business license; Enterprise is custom and the Community Edition is free to self-host. You pay per full workflow run, not per step, and test runs, sub-workflows and error workflows do not count. Self-hosting is cheaper once your server plus maintenance time costs less than the Cloud tier you would need.
n8n pricing in 2026: Cloud Starter is €20 a month billed annually (€24 monthly) for 2,500 executions, Pro is €50 billed annually (€60 monthly) for 10,000, the self-hosted Business license is €667 a month billed annually, and Enterprise is quoted. The self-hosted Community Edition is free, and every paid plan bills per full workflow run, not per step.
Prices and limits checked October 2026 on the n8n pricing page (including its monthly billing toggle), the n8n docs on how executions count and choosing Cloud or self-hosted, and the help center article on execution limits. Prices are in euros and exclude VAT; n8n changes them often, so re-check before you renew.
This page is the cost side of our n8n hub. It gives you the plan table, the counting rules that decide which tier you actually need, a worked execution estimate you can copy, and the point where running n8n on your own server becomes cheaper than paying n8n to host it.
What are the n8n cloud plans?
All plans include unlimited users, unlimited workflows and every integration. What changes is the monthly execution quota, how many executions run at once, how long history is kept, and team features such as projects, SSO and Git version control.
| Plan | Price (checked October 2026) | Executions / month | Concurrent runs | Fits |
|---|---|---|---|---|
| Starter (Cloud) | €20/mo billed annually, €24 billed monthly | 2,500 | 5 | Solo builders with a few production workflows |
| Pro 10k (Cloud) | €50/mo billed annually, €60 billed monthly | 10,000 | 20 | Small teams; adds admin roles, insights, workflow history |
| Pro 50k (Cloud) | €145 billed monthly | 50,000 | 50 | Busy chat agents and webhook-heavy workloads |
| Business (self-hosted license) | €667/mo billed annually, €800 billed monthly | 40,000 | Your server | Companies under 100 staff needing SSO, Git, environments |
| Enterprise | Custom | Custom | 200+ | Compliance, log streaming, SLA support |
| Community Edition | Free (you pay for hosting) | Unlimited | Your server | Technical users who can run and patch a server |
Two details trip people up. First, Business is not a bigger Cloud plan: it is a license key you apply to your own self-hosted instance, and n8n's FAQ says that key pings their license server daily with your production execution count. Business overages are billed at €4,000 per extra bucket of 300,000 executions. Second, the Cloud trial lasts 14 days with 1,000 executions, five concurrent runs and a 180-second execution timeout, so a trial is not a fair load test of a busy workflow.
Annual billing saves about 17%. Business is a self-hosted license, so server costs come on top. Source: n8n.io/pricing, checked October 2026
What is the n8n cost per execution, and what counts?
n8n defines an execution as one run of a workflow from start to finish. A workflow with three nodes and one with forty cost the same per run. The docs then split runs into ones that count and ones that do not, and this is where most estimates go wrong.
- Schedule Trigger: every time it fires, whatever the outcome
- Webhook Trigger: every inbound request, even an empty {} body
- Polling triggers: only polls that find new data
- Each chat agent turn (one message is one execution)
- Manual runs from the editor while building
- Sub-workflows called with Execute Sub-workflow
- Runs of your error workflow
- Polls that return nothing, and malformed webhook requests
- Executions that fail
Divide the plan price by its quota and the per-execution price falls as you move up: about €0.008 on Starter billed annually, €0.005 on Pro 10k, and roughly €0.0024 on Pro 50k billed annually. Those are ceilings, because unused executions do not roll over. The practical lesson is architectural: move shared logic into sub-workflows and keep your error workflow separate, since neither adds to your bill.
When you hit the limit on Cloud, nothing is queued. Each new run fails at once with "Execution limit reached", triggers keep firing, and runs that failed stay failed after you upgrade. Quotas reset on the first of each calendar month, not on your billing date. n8n will reset the count once per customer on Trial, Starter and Pro-1, and Pro-2 monthly subscribers can add executions up to 500,000 a month themselves.
n8n pricing calculator: how many executions do you need?
You do not need a tool to size your plan; you need an honest list of triggers. This is the method n8n's own FAQ suggests, turned into steps.
- 1List every production workflow
Only the ones that will run on a trigger. Ignore anything you only run by hand.
- 2Count schedule runs
Daily is 30 or 31 a month. Every 5 minutes is about 8,600 to 8,900 a month on its own.
- 3Count webhook events
Daily form submissions, payments or messages, times 30.
- 4Count chat turns
Conversations per week times average messages per conversation, times about 4.3.
- 5Add polling with new data only
Estimate how often the poll actually finds something, not how often it checks.
- 6Add 20% headroom and pick the tier
Over-limit runs fail rather than queue, so size for your busiest month.
Here is a worked example. The inputs are illustrative, chosen to look like a small business running four automations; swap in your own numbers.
| Workflow | Assumption | Executions / month |
|---|---|---|
| Daily report on a Schedule Trigger | Runs once a day | 30 |
| Lead form webhook | 20 submissions a day (illustrative) | 600 |
| Support chat agent | 150 conversations a week × 6 messages (illustrative) | ≈ 3,900 |
| Google Drive polling every 15 minutes | Counts only when a new file appears: 5 a day (illustrative) | 150 |
| Error workflow and sub-workflows | Do not count toward the quota | 0 |
| Total | Add 20% headroom for spikes | ≈ 4,680 → ≈ 5,600 |
About 5,600 executions with headroom is above Starter's 2,500 and comfortably inside Pro 10k. Notice where the volume comes from: the chat agent is more than 80% of the total. If you are planning an AI agent on n8n, estimate messages, not conversations, and remember that the model calls inside the workflow are billed separately by your LLM provider; our LLM API pricing comparison covers that half of the bill.
One schedule is enough to break a small plan. A single workflow on a 5-minute schedule uses roughly 8,700 executions a month before anything else runs. If it only needs to react to changes, switch it to a webhook or a polling trigger, which counts only when it finds new data. Our n8n cron expression guide helps you pick a schedule that matches the job instead of the default.
The self-host break-even point
The Community Edition costs nothing to license and has no execution quota. Its cost is the server and your time: installing, updating, backing up and fixing it when an upgrade breaks something. The break-even rule is simple:
Self-host when monthly server cost + (maintenance hours × your hourly value) is less than the Cloud plan you would need.
Our n8n Cloud vs self-hosted decision guide walks through the setup itself. The docs list what the free edition lacks: custom variables, environments, external secrets, log streaming, projects, SSO and workflow sharing all need a paid plan, even on your own server. If your team needs those, the real comparison is Cloud Pro against a Business license plus hosting, not against free.
What the break-even number ignores
- LLM API usage inside AI nodes, billed by OpenAI, Anthropic or Google, not n8n
- Backups and a tested restore of your database and credentials
- Security patches and n8n upgrades, which can include breaking changes
- Downtime: a crashed self-hosted instance misses webhooks until you notice
- Assistant credits on Cloud (1,600 on Starter, 4,000 on Pro 10k, 9,600 on Pro 50k)
- Paid features you need even when self-hosting: SSO, projects, environments
Cloud data is stored in Frankfurt, Germany, which matters if you have EU data requirements. Self-hosted, it is wherever you put it.
Which n8n plan should you pick?
- Trial first, always. Fourteen days and 1,000 executions are enough to build your real workflows and read the execution count in the Insights dashboard before you pay anything.
- Starter fits one person running a handful of scheduled and webhook workflows under 2,500 runs a month. Its limit is five concurrent executions, so bursts of webhooks will queue.
- Pro 10k is the default for client work and small teams: 20 concurrent runs, admin roles, workflow history and three shared projects.
- Pro 50k is for chat agents and busy webhooks. Each agent turn is an execution, and 50 concurrent runs keep replies from waiting in line.
- Business makes sense when you already self-host and need SSO, Git-based version control, multiple environments or queue mode.
- Community Edition is the right answer for technical builders who want unlimited executions and accept doing their own upkeep.
Whichever you pick, set a calendar reminder for the last week of each month while you learn your real usage. Because over-limit runs fail instead of waiting, the first sign of an undersized plan is usually a customer asking why a form or a bot went quiet.
n8n pricing vs Zapier and Make
Zapier bills per task, meaning each successful action step; Make bills in credits, usually one per module action and per bundle processed; n8n bills per execution (all checked October 2026). For long workflows that branch, loop and call several APIs, per-execution billing is usually cheaper. For short, very frequent workflows, the models converge, and ease of use may matter more than price. Our n8n vs Zapier breakdown, the Make vs n8n comparison and the side-by-side pages for n8n vs Zapier and n8n vs Make carry the details.
If your automations are turning into a product people would pay for, our AI SaaS Builder program covers the next step: validating the idea, building it on Next.js and Supabase, and charging for it.
For the server side of self-hosting costs, with VPS list prices and upkeep, see what self-hosted n8n really costs.
n8n pricing: FAQ
How much does n8n cost in 2026?
n8n Cloud Starter costs €20 a month billed annually or €24 billed monthly for 2,500 executions. Pro costs €50 a month billed annually or €60 monthly for 10,000 executions, with a 50,000-execution tier above it. Business, a self-hosted license, is €667 a month billed annually. Enterprise is custom. The self-hosted Community Edition is free. Prices checked October 2026 on n8n.io/pricing.
Is n8n free?
The self-hosted Community Edition is free and n8n describes it as having almost the complete feature set. You still pay for the server and spend time on updates and backups. n8n Cloud has no permanent free plan; it has a 14-day trial with 1,000 executions and no credit card required. Features like SSO, environments and projects need a paid plan even when you self-host.
What counts as an execution in n8n?
An execution is one full run of a workflow, however many steps it has. Only production executions count: schedule triggers count every time they fire, polling triggers only when they find new data, and webhooks once per inbound request. Manual test runs, sub-workflow runs, error workflow runs, polls with no data and failed executions do not count toward the quota.
What happens when I hit my n8n execution limit?
On Cloud plans, executions over the limit are not queued. Each one fails immediately with an "Execution limit reached" error, and failed runs are not retried automatically after you upgrade. Limits reset at the start of each calendar month, not on your billing date. n8n offers a one-time courtesy reset on Trial, Starter and Pro-1, and Pro-2 monthly subscribers can add executions themselves.
When is self-hosting n8n cheaper than n8n Cloud?
Self-hosting wins when your server cost plus the value of the hours you spend maintaining it is lower than the Cloud plan you would otherwise need. On a small workload that fits Starter, a few hours of maintenance a month usually costs more than the plan. On workloads above 10,000 executions, or when you need unlimited executions, self-hosting usually comes out ahead if you can run servers.
Does n8n charge per step like Zapier?
No. n8n bills per workflow execution, so a 20-step workflow that runs once is one execution. Zapier bills per task, where each successful action step counts, and Make bills in credits, usually one per module action. That makes n8n cheaper for long workflows, but the gap shrinks for short, high-frequency workflows such as a webhook that does one thing thousands of times a day.
Free · no card needed
Start with the free Starter Kit.
A unit economics calculator from AI SaaS Builder, plus four more working files.
Automate the work. Then build the product.
AI SaaS Builder, included in All Access, covers Supabase, the Claude API and shipping a paid AI product, with the other three programs, live coaching and the private community in one subscription.
Sizing an n8n setup?
Ask in the free Discord, where members share n8n workflows and plan choices, or grab ready workflows from our template library post.