n8n Pricing (2026): Cloud Plans, Executions and Self Hosting Costs

The price list

n8n publishes its prices in euros only. The USD column converts at the European Central Bank reference rate of 1.1253 (October 16, 2026), rounded to the cent.

Plan	Billed monthly	Billed annually (per month)	Executions a month	Where it runs
Community Edition	Free	Free	No cap, your hardware	Self hosted
Starter	about $27 (€24)	about $23 (€20)	2,500	n8n Cloud
Pro 10K	about $68 (€60)	about $56 (€50)	10,000	n8n Cloud
Pro 50K	about $163 (€145)	about $136 (€120.83)	50,000	n8n Cloud
Business	about $900 (€800)	about $751 (€667)	40,000, tiers to 500,000	Self hosted only
Start-up Plan	50% off Business	about $375 (€333)	40,000	Self hosted only
Enterprise	Quote only	Quote only	Custom	Cloud or self hosted

Annual billing saves 17%, according to the pricing page. Every plan includes unlimited users and unlimited workflows, so you never pay per seat.

Business scales by execution volume. The tiers listed in the page's plan selector, billed monthly: 50,000 for about $1,108 (€985,) 75,000 for about $1,638 (€1,456,) 100,000 for about $1,942 (€1,726,) 150,000 for about $2,463 (€2,189,) 200,000 for about $2,970 (€2,639,) 300,000 for about $3,952 (€3,512,) 400,000 for about $4,930 (€4,381) and 500,000 for about $5,902 (€5,245).

What each limit means in practice

Executions. One execution is one run of a whole workflow. It does not matter whether the workflow has 3 steps or 30, or whether it moves one row or a thousand. This is the core difference from Zapier, which counts every action as a task, and Make, which counts every module run as a credit. One catch for AI builds: n8n counts each agent turn as one execution, so a support chat with 6 back and forth messages uses 6.

Concurrent executions. This is how many runs can happen at the same moment. Starter allows 5, Pro allows 20 on the 10,000 tier and 50 on the 50,000 tier, Business 20 to 50, and Enterprise 200 or more. When you hit the cap, new runs wait in a queue rather than fail. A busy chatbot hits this before it hits the monthly quota.

Shared projects. Projects let teams share workflows and credentials with permissions. Starter has 1, Pro 3, Business 6 and Enterprise unlimited.

Log retention and insights. Starter keeps execution logs for 7 days with limited insights. Pro keeps 30 days with 7 days of insights data. Business keeps 30 days with a 14 to 30 day insights dashboard. Enterprise keeps 365 days or more.

AI Assistant credits. These power n8n's built in helper that writes and fixes workflows for you. Starter includes 1,600 a month and Pro up to 9,600. They refresh monthly and do not roll over. They are not the same as the model calls inside your workflows.

Gateway credits. On Starter and Pro cloud plans, you can call OpenAI, Anthropic and Google Gemini models, plus tools such as Brave Search and Firecrawl, without your own API keys. n8n bills each request at the rates on its service pricing page, and top ups expire 12 months after purchase. If you bring your own API keys, you pay the provider directly instead.

Cloud vs self hosted

The Community Edition is the same core product as every paid plan. Without a license key, a self hosted instance runs as Community. You get unlimited executions and workflows, plus code nodes in JavaScript and Python, but no SSO, no shared projects with role based access, no version control environments and only forum support.

Business adds those team features to your own server through a license key. Two details matter before you buy it. It is self hosted only, since the cloud version is on a waitlist. And the license pings n8n's server daily to count production executions, so an air gapped install needs Enterprise.

Cloud makes sense when you want n8n to handle uptime, updates and backups. Self hosting makes sense when volume is high, data must stay on your infrastructure, or you need more than 50,000 executions without paying Business prices.

What self hosting costs

The software is free. The bill is the server plus your time. Here is a minimal production setup on DigitalOcean, as listed in October 2026:

Item	Monthly cost
Basic droplet, 2 GB RAM, 1 vCPU	$12.00
Managed PostgreSQL, 1 GB RAM, 1 vCPU, 10 GB storage	$15.15
Total	$27.15

A 4 GB, 2 vCPU droplet is $24, which brings the total to $39.15 if your workflows process large files or run many agents at once. You can skip the managed database and run PostgreSQL on the same server to stay at $12, but then backups are on you. n8n's docs recommend Docker Compose for production and note that npm installs are deprecated from n8n 3.0.

The hidden cost is maintenance. You patch the server, update n8n, rotate credentials, monitor memory and restore from backup when something breaks. At even two hours a month of an engineer's time, self hosting is not truly free. It wins on price once your volume would push you past the 10,000 execution Pro tier.

Worked examples

All examples assume cloud billing monthly unless noted, with USD at the October 16, 2026 ECB rate.

1. Lead routing for a small agency. A webhook catches 50 form leads a day, enriches each one, writes to the CRM and posts to Slack: 6 steps. That is 1,500 executions a month, which fits Starter at about $27 (€24). The same workflow uses 9,000 tasks on Zapier, which lands in the 10,000 task Professional tier at $193.50 a month billed monthly, and about 9,000 credits on Make, which fits Core at $16.

2. A schedule that polls every 5 minutes. 12 runs an hour, 24 hours a day, for 30 days is 8,640 executions, even when nothing changed. That needs Pro 10K at about $68 (€60). Polling every minute is 43,200 runs and needs Pro 50K at about $163 (€145). Switch to a webhook trigger where the source app supports one and the same job can fit Starter.

3. Website support agent. 100 chats a day with an average of 5 agent turns is 15,000 executions a month. That is over the 10,000 tier, so you need Pro 50K at about $163 (€145), or about $136 (€120.83) billed annually. Model calls are extra, through Gateway credits or your own API keys.

4. Growing team that needs SSO. Ten people, 30,000 executions a month, single sign on required. SSO starts at Business, so the answer is about $751 (€667) a month billed annually, or about $375 (€333) on the Start-up Plan if you qualify. That is the steepest step on the price list: from about $163 (€145) on Pro to about $900 (€800) monthly on Business.

5. High volume operations, 300,000 executions. Business at the 300,000 tier is about $3,952 (€3,512) a month billed monthly. Self hosting the Community Edition on a $39.15 setup runs the same volume with no license fee, if you can live without SSO and shared projects.

6. Solo builder on a budget. Self host on a $12 droplet with SQLite or a local PostgreSQL. Total $12 a month, unlimited executions, and you own every update.

Hidden costs and overages

• Overage on Business. Extra executions come in buckets of 300,000 for about $4,501 (€4,000), invoiced automatically 45 days after the overage if you do not upgrade.
• Cloud overage. n8n says workflows keep running when you pass your quota but overage charges may apply, and its sales team contacts you near the limit.
• Exchange rate. The price is fixed in euros, so a weaker dollar raises your cost. A 5% move in the rate changes a about $751 (€667) Business bill by about $37 a month.
• Model usage. Execution prices do not include AI model calls. Budget separately for Gateway credits or your OpenAI, Anthropic or Gemini bill.
• Trial deletion. If you do not upgrade by the end of the 14 day trial, n8n deletes the workspace. You have 90 days to download your workflows.

How n8n compares on price

Platform	Entry paid plan, billed monthly	What the entry plan includes	Billing unit
n8n	about $27 (€24)	2,500 executions	One per workflow run
Make	$16	10,000 credits	One per module run, AI steps can cost more
Zapier	$29.99	750 tasks	One per action step
Gumloop	$37	20,000 credits	Model and tool usage at list price plus 8%
Lindy	$49.99 per user	5,000 credits	Per task, 2 to 2,500 credits

Make wins for short, low volume workflows. n8n pulls ahead as workflows get longer, and self hosting removes the subscription entirely. Zapier costs more per run but has 9,000+ app integrations against n8n's 500+. See every agent builder's unit price in the AI Price Index, compare head to head in n8n vs Zapier and n8n vs Make, or browse n8n alternatives.

How to keep the bill down

• Use webhooks, not polling. Scheduled checks burn executions even when nothing happened.
• Batch inside one run. Loop over 100 records in one execution instead of triggering 100 runs.
• Pay annually once your volume is stable. It saves 17%, about $54 (€48) a year on Starter and about $135 (€120) on Pro 10K.
• Self host past 10,000 runs. If your team does not need SSO, a $27.15 server beats Pro 50K at about $163 (€145).
• Check the Start-up Plan. Under 20 employees and under about $5.63 (€5)M raised gets Business at half price.

Is n8n worth the money

For technical teams running multi step or AI agent workflows, n8n is the best value in the category: per run billing, unlimited users on every plan and a free self hosted edition. The weak spot is the jump from Pro to Business, where SSO alone moves the bill from about $163 (€145) to about $751 (€667) a month or more, and the business plan only runs on your own servers. Read the full n8n review for features, ratings and who should skip it.