The short version
All figures captured from the live n8n pricing page on 2026-09-16 (proxied browser; the Annually and Monthly controls were both exercised, and the full feature matrix below the cards was captured separately). The page presents every price in euro (€) — with our exit point the site shows no currency selector anywhere in the DOM, so we quote € exactly as printed without converting.
| Plan | Yearly (€/mo) | Monthly (€/mo) | Workflow executions | Hosting |
|---|---|---|---|---|
| Starter | €20 | €24 | 2,500/mo | Hosted by n8n |
| Pro | €50 | €60 | 10,000/mo | Hosted by n8n |
| Business | €667 | €800 | 40,000/mo | Self-hosted |
| Enterprise | Custom | Custom | Custom | Hosted or self-hosted |
| Community Edition | $0 | $0 | Unlimited (your server) | Self-hosted, Sustainable Use License (fair-code, not open source) |
The monthly price is a flat +20% on every tier (€20→24, €50→60, €667→800); the toggle label says “Save 17%”, which is the yearly-direction discount computed the other way (20/24 = 16.7%) — both statements are the same deal. Billing is per execution, not per step: the page states twice that pricing is “based on monthly workflow executions, regardless of complexity” and “Pay for full executions, not for each step”, so a 10-step workflow costs the same one execution a 2-step one does — the opposite axis from per-task tools like Zapier. This page carries the full plan mechanics, the FAQ clauses below the cards, and the numbers the four cards don’t print; the standalone 2026-09-16 terms check formerly at /saas/n8n-review/ was merged into it on 2026-10-20 (one pricing intent, one URL).
What the cards omit: the limits matrix
The pricing cards lead with one number — workflow executions. The full comparison matrix below them (captured in evidence image 3) publishes the limits that actually decide whether a plan works in production:
| Starter | Pro | Enterprise | |
|---|---|---|---|
| Max saved executions | 2,500 | 25,000 | 50,000 |
| Saved-execution storage | 2.5 GB | 25 GB | 50 GB |
| Max execution log retention | 7 days | 30 days | Unlimited |
| Concurrent executions | 5 | 20 | 200+ |
| Workflow history (max days) | 1 | 5 | 365+ |
| Shared projects | 1 | 3 | Unlimited |
| AI Assistant credits | 2,300/mo | Up to 13,700/mo | N/A (sized) |
Two consequences worth flagging:
- The Starter card says “2.5k workflow executions”, and the matrix separately caps “max saved executions” at the same 2,500. On Starter the execution history you can inspect is therefore only 7 days or 2,500 runs, whichever comes first — debugging a workflow that ran last week means upgrading to Pro (30 days) or Enterprise (unlimited retention).
- Five concurrent executions on Starter is the quiet throughput ceiling: many short runs queued simultaneously will wait, regardless of how few of your 2,500 monthly executions you’ve used.
Unit prices: what an execution costs on each tier
Dividing the monthly subscription by the included execution quota (yearly state, simple arithmetic on the page’s own numbers, not a vendor quote):
| Plan | € / execution (yearly) | € / execution (monthly) |
|---|---|---|
| Starter | €0.0080 | €0.0096 |
| Pro | €0.0050 | €0.0060 |
| Business (self-hosted, 40k) | €0.0167 | €0.0200 |
Two readings:
- Pro is the cheapest execution on cloud by a wide margin (€0.005 vs €0.008) — moving from Starter to Pro doubles the per-euro quota and unlocks 20 concurrency and 30-day history.
- Business at €0.0167 per included execution is more than 3× the Pro unit price. You are not buying cheaper volume at Business; you are buying self-hosting, SSO/SAML/LDAP, environments, Git version control and scaling options. Judging Business on per-execution price misses the point — see the overage clause in the next section before signing a Business contract.
Rules the plan cards don’t print (from the page’s own FAQ, captured 2026-09-16)
- Overage on self-hosted Business/Enterprise: exceeding your quota does not stop workflows, “but overage charges may apply.” The sales team contacts you near quota; if there’s “no response or upgrade path, the overages if any will be automatically invoiced 45 days after they have occurred. For the Business plan, overage prices are 4,000 EUR for extra buckets of 300,000 executions.” A €4,000 invoice arriving 45 days after an unnoticed quota breach is the single most important sentence on this page — set your own quota alerts.
- Cloud saved-execution caps are history limits, not run limits. Reaching the saved-execution limit (2.5k/25k/50k by tier) deletes the oldest history; “Reaching a limit does not stop your workflows from running.” What a Cloud plan does at its monthly execution quota is not stated in the FAQ — not verified.
- AI Assistant credits are a hard monthly allowance (2,300 Starter, 5,700 or 13,700 Pro by plan size, 800 on trial): no rollover, and “it’s not yet possible to buy additional credits” — the only route to more is a higher tier. Self-hosted AI Assistant is “coming soon” with bring-your-own API key and no credit limits.
- Trial / credit-card rules: 14-day trial gives all Pro features with Starter limits (5
concurrent, 2,500 executions) and 800 AI credits. No credit card required on Starter/Pro
trials — a card is required only for the Business trial. The signup form we reached
(
app.n8n.cloud/register) shows a single company-email field and no card inputs. - Refunds: no pro-rata. Monthly cancels at the end of the billing cycle; annual cancels at the end of the annual term.
- Startup discount: companies under 20 employees can get 50% off Business (separate startup-plan page).
- Data residency: hosted plans store data in the EU (Frankfurt, Germany) — relevant if your compliance story is US-centric.
- The Business license is a license key that pings n8n’s license server daily to stay active; the Community Edition needs no license and no key.
“Unlimited users” vs metered executions
Every paid plan prints “All plans include unlimited users & workflows” — the meter is purely the execution count. That is the exact opposite axis from seat-priced automation tools, and it interacts with two other published facts:
- Workflows and active workflows are unlimited even on Starter; the limits are runs, concurrency, projects and history.
- AI credits are a separate, non-rollover pool (2,300/mo on Starter, up to 13,700/mo on Pro), and the FAQ states it is currently not possible to buy more — if your automation is AI-heavy, the AI pool is the binding constraint before the execution quota is.
Where the free option fits in the price ladder
There is no free cloud tier. The zero-price path is the Community Edition — a standard self-hosted build on GitHub (204,346 stars on the repository at the 2026-09-16 capture), covering all core features, all integrations and unlimited executions on your own infrastructure. The pricing matrix itself routes three capabilities — custom nodes, running bash scripts and a few developer controls — to “Available in self-hosted” on paid cloud plans. In other words the ladder is: free self-hosted (your DevOps) → €20 hosted cloud → €667 self-hosted with enterprise controls → custom Enterprise; cloud trials require no credit card, the Business trial does.
Prices and limits captured from n8n.io/pricing on 2026-09-16 (Annually and Monthly states plus the feature matrix, all three in the evidence panel). Unit-price columns are our arithmetic on the page’s own numbers. Quoted euro figures are the page’s only currency; the site offered no USD switch in this capture. See our testing methodology for how we verify claims.
Related guides
More of our n8n guides:
