RecipeAgents / Guides
Agent recipe guiden8n pricing: is n8n free?
Is n8n free? Yes, if you self-host the Community Edition: no licence fee and no execution quota, but you pay for the server and the upkeep. n8n Cloud has no free plan, only a trial; paid plans start at €20 a month. n8n charges per execution: one full run of a workflow, however many nodes it has. On n8n.io/pricing on Oct 26, 2026, Cloud Starter was €20 a month billed annually (€24 monthly) for 2,500 executions, Pro €50 (€60 monthly) for 10,000, and the self-hosted Business plan €667 a month billed annually (€800 monthly) for 40,000. The self-hosted Community Edition is free and has no execution quota.
The price list is the easy part. What decides your plan is how your triggers count. We went through the triggers in our 25 downloadable n8n workflows: 5 run on a weekly schedule and use about 22 executions a month between them, 1 runs every weekday (about 22 on its own), 5 fire on events, 1 is manual only, and 13 poll another app, 9 of them every minute. Those 9 are where a quota can disappear, and the docs and a recent forum thread disagree on why.
n8n cost calculator: executions and plan
n8n cost calculator
Add each active workflow and how it starts. The three rows below are example inputs: change them to yours. You get two numbers: executions as n8n's docs count them, and the worst case if every poll counts. Runs in your browser.
Assumptions behind the numbers
- Prices in euros from the plan data on n8n.io/pricing, read Oct 26, 2026. Cloud: Starter 2,500 executions (€24 monthly, €240 a year), Pro 10,000 (€60, €600 a year), Pro 50,000 (€145, €1,450 a year). Self-hosted Business starts at 40,000 (€800, €8,000 a year) with tiers up to 500,000. Above 50,000 on cloud you're talking to n8n's sales team.
- One execution is one production run of a whole workflow, however many nodes it has. Manual test runs, sub-workflows and error workflows don't count (n8n docs).
- A month is 30 days: every minute is 43,200 runs, every 5 minutes 8,640 (n8n's own FAQ says about 8,600 to 8,900), every 15 minutes 2,880, hourly 720, daily 30, weekly 4.3.
- Polling triggers (Gmail, Google Drive, Sheets, RSS): the docs say a poll that finds nothing isn't an execution, so the docs number counts the items you expect, capped at one per poll. A May 2026 forum thread reports empty polls being counted, so the worst case counts every poll. Check your Usage page after a week.
- Not included: AI model or API costs, which the model provider bills; server costs for self-hosting; VAT. The free Community Edition has no execution quota.
- Limits that can force a bigger plan regardless of volume: Starter allows 5 runs at once and 5 minutes per run; Pro 20 at once (50 on the 50k tier) and 40 minutes.
Step by step
- List every active workflow and its trigger type: schedule, webhook or form, polling (Gmail, Google Drive, Google Sheets, RSS) or manual.
- Schedules: count the fires per month. Weekly is about 4.3, daily 30, hourly 720, every 5 minutes 8,640.
- Webhooks and forms: count the events you expect, for example leads per month. Each request is one execution.
- Polling triggers: count the items you expect, but also write down the ceiling, one execution per poll. Every minute is 43,200 a month, every 15 minutes 2,880, hourly 720.
- Add it up, leave room for the polling ceiling, then pick the plan. After a week, compare with the real number on your Usage page and the Executions list.
n8n plans and prices, checked Oct 26, 2026
Prices as n8n.io/pricing showed them to us in euros. The page shows the same Starter and Pro numbers in dollars to US visitors, and Business at $800 billed annually; our n8n vs Zapier and n8n vs Make pages use the dollar list.
| Plan | Billed annually | Billed monthly | Executions a month | Limits worth knowing |
|---|---|---|---|---|
| Community (self-hosted) | €0 | €0 | No quota | You run and pay for the server |
| Starter (cloud) | €20/mo | €24 | 2,500 | 5 runs at once, 5-minute max per run, 1 shared project |
| Pro (cloud) | €50/mo | €60 | 10,000 (50,000 for €145/mo) | 20 runs at once (50 on the 50k tier), 40-minute max per run |
| Business (self-hosted) | €667/mo | €800 | 40,000, tiers up to 500,000 | SSO, Git version control, queue mode |
| Enterprise | Contact sales | Custom | Cloud or self-hosted | |
Two more lines from the same page. The free cloud trial gives 1,000 executions with a 180-second timeout per run. Companies under 20 employees can apply for the Start-up plan, Business at half price (€333 a month billed annually). Business and Enterprise keep running past the quota and are invoiced later, at €4,000 per extra 300,000 executions on Business.
How each trigger counts, and what our 25 workflows use
The rules are in n8n's Understand executions docs, read Oct 26, 2026: only production runs count. Manual test runs, sub-workflows and error workflows don't. Here they are applied to the 25 n8n JSON files in our n8n templates library:
| Trigger | Our workflows | How n8n counts it | Executions a month |
|---|---|---|---|
| Schedule, weekly | 5 (e.g. weekly ops digest) | Every fire, whatever the outcome | About 4.3 each, 22 for all five |
| Schedule, weekdays | 1 (Slack channel digest) | Every fire, even on a quiet day | About 22 |
| Webhook, form, Calendly, Telegram | 5 (e.g. new lead to CRM note) | One per inbound request, even an empty one | Your number of events |
| Polling, every hour | 4 (RSS, 2 Google Sheets, 1 Gmail) | Only polls that find new data | Up to 720 each |
| Polling, every minute | 9 (5 Gmail, 4 Google Drive) | Only polls that find new data | Up to 43,200 each |
| Manual | 1 (image alt text) | Manual runs don't count | 0 |
Node count doesn't matter: 21 of our 25 workflows have three nodes, two have four, the Slack channel digest has five and the Telegram AI agent has eleven, and each run is one execution whether it has 3 nodes or 30. The AI step inside it isn't billed separately by n8n either.
Most public workflows lean the same way. In our study of 13,574 public n8n templates (snapshot 2026-10-20), 4,567 (33.6%) use a Schedule Trigger, 2,676 (19.7%) a webhook, 1,677 (12.4%) a form trigger and 641 (4.7%) an email trigger, while 2,906 (21.4%) have a manual trigger, which costs nothing in production. A workflow can have more than one trigger, so the shares overlap.
The polling catch: docs say empty polls are free, one forum thread says otherwise
Per the docs, a Gmail or Google Drive trigger that polls every minute and finds nothing doesn't use an execution. So our angry-email flag on an inbox that gets 15 matching emails a working day costs about 330 executions a month, fewer when two arrive in the same minute. That fits Starter with room to spare.
A May 2026 thread on the n8n forum, Gmail Trigger Error, reports the opposite on n8n Cloud 2.20.7: a Gmail Trigger polling for unread mail logged runs, as errors, when no mail was found, and they counted toward the quota. The poster says this didn't happen on 2.16. If empty polls count, one every-minute trigger costs 43,200 executions a month, 17 times Starter's 2,500 and over four times Pro's 10,000.
So check before you trust either version. Turn the workflow on, wait a day, then open the Executions list. If you see runs with no new email in them, change Poll Times on the trigger to every 15 minutes (2,880 a month at most) or every hour (720). For a bill reminder or invoice filer, an hour's delay rarely matters. Our files ship with every-minute polling because it feels instant when you test; it's the first setting to change on a Starter plan.
What the plan price leaves out
Model costs. Unless you use n8n's own AI credits, every OpenAI, Claude or Gemini call in a workflow is billed by the model provider on your own API key, separate from n8n. For a per-run comparison that includes the AI step, see n8n vs Zapier and n8n vs Make; on the n8n side the AI step is still one execution.
Self-hosting isn't zero. The Community Edition has no execution quota, but you pay for the server, updates, backups and the hours you spend on them, and it's under n8n's Sustainable Use License, which covers your own internal use. The paid self-hosted plan is Business, from €667 a month, so the jump from free to paid is steeper on self-hosted than on cloud.
Run length. Starter stops a run after 5 minutes. Our recipes finish in seconds, but a workflow that loops over a big sheet or waits on a slow model can hit that limit before it hits the execution quota.
Before you use it
Every recipe on RecipeAgents follows the same shape: a trigger, the tools, the steps, and the exact prompt. Start with the human-approval version. Once you have watched a few dozen runs and trust the output, decide which steps can run on their own. Model and tool names change often, so check current pricing and features in each tool before you build.
Frequently asked questions
- Is it free to use n8n?
- The self-hosted Community Edition is free and has no execution quota; you pay for the server. n8n Cloud has a free trial with 1,000 executions, then plans start at €20 a month billed annually (€24 monthly) for 2,500 executions, checked Oct 26, 2026.
- How much does an n8n monthly subscription cost?
- On Oct 26, 2026, billed monthly: Starter €24 for 2,500 executions, Pro €60 for 10,000, and self-hosted Business €800 for 40,000. Billed annually it's €20, €50 and €667 a month.
- Why is n8n so expensive?
- For most small setups it isn't, because a run with 30 nodes costs the same one execution as a run with 3. It gets expensive when frequent polling or schedules burn executions without doing anything useful, or when you need a self-hosted paid plan, which starts at Business.
- Do failed runs count as executions?
- For a Schedule Trigger, yes: the docs say every fire counts regardless of outcome. Manual test runs and error workflows don't count.
- Does a 20-node workflow cost more than a 3-node one?
- No. n8n counts the run, not the nodes or the items in it, so one execution either way.