n8n Pricing: n8n Cloud Pricing, n8n Cost, Self-Hosted Pricing, the Starter Plan and n8n Enterprise Pricing in US Dollars

n8n sells four paid plans. Starter and Pro are n8n Cloud plans, hosted by n8n in its own data centers. Business is a license for a copy of n8n you run yourself. Enterprise can be either and is priced by sales. Every plan includes unlimited users, unlimited workflows and every integration, which is unusual in this category: you never pay for another seat when a colleague joins, and you never pay extra for a connector.

What you pay for is volume. Each plan carries a monthly quota of workflow executions, and the price rises with the quota. On monthly billing Starter is 24 EUR for 2,500 executions, Pro is 60 EUR for 10,000, and Business is 800 EUR for 40,000. Annual billing is priced at exactly ten months of the monthly rate, so the same three plans work out to 20 EUR, 50 EUR and 666.67 EUR a month when paid a year upfront. We read these off the plan data behind n8n.io/pricing on September 11, 2026, where the annual and monthly figures for every size are published side by side: 240 and 24 for Starter, 600 and 60 for Pro, 8,000 and 800 for Business.

The fastest way to read the list is by what changes as you move up. Starter to Pro buys four times the executions, four times the concurrency and admin features. Pro to Business buys the features a company's IT team asks for, SSO, Git version control and separate environments, and moves you off n8n's servers onto your own. Business to Enterprise buys dedicated support with an SLA, log streaming, external secret stores and invoice billing on n8n's hosted cloud.

n8n is a Berlin company, and its pricing page lists every plan in euros with no currency switch. That matters more than it looks, because many of the guides ranking for n8n pricing quote n8n in dollars, usually by swapping the euro sign for a dollar sign. At the European Central Bank reference rate of $1.1616 per euro on September 10, 2026, the real numbers are 16% higher than those guides suggest:

How you pay changes the final number. Starter and Pro are self-serve and billed by card in euros, so your bank converts the charge on the day it posts and your statement will drift with the exchange rate from month to month. Many US business cards also add a foreign transaction fee, commonly up to 3%, unless the card waives it, so check the card before you put a recurring euro charge on it. The annual Business plan is different: n8n's pricing FAQ offers payment by invoice and wire transfer, with one self-service contract form for euro payment and a separate one for dollar payment. If your finance team wants a fixed dollar invoice, annual Business and Enterprise are the plans that offer one.

The two Cloud plans share the same editor, the same integrations and unlimited users. The differences are capacity and administration, and n8n publishes each one:

Concurrency is the limit people notice first. Starter runs five executions at once and queues the rest, which is fine for a scheduled report and tight for a chatbot or a webhook that fires in bursts. If your workflows are triggered by customers rather than by a clock, budget for Pro even when the execution count says Starter would do.

AI Assistant credits power n8n's built-in assistant for building and editing workflows. They are not the model costs your AI agent workflows incur when they call OpenAI or Anthropic, which you pay to those vendors on your own API key. n8n's FAQ says the monthly credit balance does not roll over and that additional credits cannot yet be bought, so the only way to get more is a bigger plan.

The pricing page shows Pro at 10,000 executions, but the plan has a second size in its dropdown: 50,000 executions for 145 EUR a month, or 1,450 EUR a year (about $168.43 monthly, or $1,684.32 a year). Five times the executions for 2.4 times the price.

That makes the larger Pro size the cheapest executions on n8n's entire price list. Here is every published size divided into a cost per run, on monthly billing and converted at $1.1616:

A team running up to 50,000 workflow runs a month that does not need SSO or Git gets the lowest unit price n8n sells on Pro, at a fraction of a cent a run. Annual billing takes a further sixth off every figure in the table.

Self-hosting n8n comes in three license levels. The Community Edition is published on GitHub with no license fee. The Business plan is a paid license key for your own instance, and n8n's Start-up plan page states it plainly: "The Business plan is currently available only for self-hosted deployments." Enterprise can run on your servers too.

The Business entry price of 800 EUR a month covers 40,000 executions. Above that, n8n publishes a ladder of larger sizes that only shows up in the plan selector:

On top of the license you run the servers, the database, backups, upgrades and monitoring, which is real engineering time even when the hosting bill is small. Business includes six shared projects, SSO with SAML and LDAP, Git version control, separate environments, scaling options and 30 days of insights, but support is the community forum: n8n's FAQ says dedicated support is only in Enterprise.

Put the two tables above next to each other and the ladder runs backwards. Business at 40,000 executions costs 0.02 EUR a run. Pro at 50,000 costs 0.0029 EUR a run, and n8n hosts it for you. Even the largest Business size, 500,000 executions, is 0.0105 EUR a run, more than three times the larger Pro size.

The reason is that Business is not priced on executions. It is priced on the controls a company needs once more than a handful of people depend on the workflows: single sign-on, version control in Git, dev and production environments, and the scaling options n8n reserves for licensed instances. n8n's own FAQ makes the logic explicit when it answers why self-hosters pay by usage at all: its pricing is tied to "the value it brings," not to what it costs n8n to run.

So the buying question is not how many executions you need. It is whether your security review requires SSO and whether your engineers want workflows in Git. If neither is true, Pro on n8n Cloud is the cheaper plan at every volume up to 50,000 runs a month. If either is true, you are buying Business for the controls, and the execution quota comes with it.

n8n only publishes an overage rate for the licensed plans. For Business, its FAQ gives it in one line: overage is charged at 4,000 EUR for each extra bucket of 300,000 executions, which is 0.0133 EUR (about $0.0155) per execution if you use the whole bucket, and far more if you only spill over by a few thousand runs. Workflows keep running when you pass the quota. n8n's sales team contacts you as you approach it, annual Business subscribers get a call near 80% of the annual quota, and if there is no upgrade the overage is invoiced automatically 45 days after it occurs.

Two mechanics matter for self-hosted budgets. First, a Business or Enterprise license key can be applied to an unlimited number of instances, and the usage of every instance counts against one quota, so a separate staging server spends from the same pool as production. Second, the license key pings n8n's license server daily and reports production execution counts, so usage is metered even on your own hardware.

Saved-execution limits are a different thing and cost nothing. Each plan keeps a fixed number of past executions for debugging and deletes the oldest when it hits the cap, but n8n says reaching that limit "does not stop your workflows from running."

n8n does not publish an Enterprise price. The pricing page says "Contact Sales" and lists a custom number of workflow executions. What the contract adds over Business is published: unlimited shared projects, 200 or more concurrent executions, 365 days of insights, external secret store integration, log streaming to tools such as Datadog, extended data retention, dedicated support with an SLA and invoice billing. Enterprise can be hosted by n8n or self-hosted, which makes it the only way to get SSO on n8n's own cloud.

US buyers should ask one question early. n8n's FAQ says that for hosted plans, data is stored in the EU, on servers in Frankfurt, Germany. If your security questionnaire requires US data residency, the only hosting location n8n publishes for its cloud is Frankfurt, so confirm any other region with sales in writing, and the route n8n documents for full control is a self-hosted Business or Enterprise instance in your own US cloud account.

Companies with fewer than 20 employees can apply for n8n's Start-up plan, which is the Business plan at 50% off: 333 EUR a month billed annually (about $386.81 a month, or about $4,642 a year), with the same 40,000 executions and the same self-hosted-only deployment. It keeps SSO, version control and environments, so a small technical team that needs those controls can have them at half the list price.

The n8n Cloud free trial runs 14 days with Pro features, capped at 1,000 executions, five concurrent executions, a 180 second execution timeout and one team project, and it needs no credit card. A Business trial also lasts 14 days but requires a card. If you do not upgrade, n8n deletes the trial workspace, and you have 90 days to download your workflows.

n8n's own sizing advice is to count triggers, not steps, and it is the right method. Every time a workflow starts and finishes, that is one execution. A few worked numbers make the plans easy to pick:

The chatbot row is the one teams underestimate. n8n counts each message as a new execution, so a conversation of six messages is six executions, not one. AI agent workflows built on n8n usually sit behind a chat or a webhook, which is why they climb through the plan sizes faster than scheduled automations do.

n8n is one of the cheapest ways on the market to run automation you design yourself. At 0.0029 EUR a run on the larger Pro size, the platform bill is rarely the expensive part. The expensive part is the person who designs the workflow, wires the credentials, writes the prompts, handles the edge cases and fixes it when an upstream API changes, and none of that shows up on n8n's price list.

That trade is worth making when the logic is genuinely yours. It is a poor trade for a job other companies already run the same way: qualifying inbound leads, triaging a support queue, chasing unpaid invoices or reconciling statements. For those, a ready-made agent that already does the job is priced on the job, and it starts working without a build.

So price both before you commit. If you are also weighing Zapier, our page on AI agents for Zapier covers its separate agent meter, and n8n alternatives compared by price lines n8n up against the other workflow and agent builders. Our AI agent platforms comparison puts published seat and consumption rates from the big enterprise vendors side by side, and the agents above show what the finished job costs when someone else has already built it.