Free
Token calculator

n8n pricing: Cloud plans, executions and self-hosting

n8n Cloud starts at €20 a month billed yearly for 2,500 executions; Pro is €50 for 10,000. Self-hosting is free. What each plan, credit and limit costs.

By · Updated

Quick comparison

ProviderFromStarting planFree tierOur rating
n8n CloudChecked
See sitePrices being checkedFree trial only (self-hosting is free)Not ratedVisit
ZapierChecked
$19.99/moProfessional (lowest task tier, 750 tasks)750 tasks/month at the lowest tier (page data: 'Pro 750'); multi-step Zaps; unlimited premium apps; webhooks; AI fields; email support (live chat from the 2,000-task tier)100 tasks/month, two-step ZapsNot ratedVisit
MakeChecked
$12/moCore10,000 credits/month; unlimited active scenarios; scheduling down to the minute1,000 credits/monthNot ratedVisit
Hetzner CloudChecked
$6.49/moCX23 (Cost-Optimized, Intel/AMD)2 vCPU · 4 GB RAM · 40 GB NVMeNoNot ratedVisit
HostingerChecked
$2.49/moPremium (web hosting)20 GB SSDNo (30-day money-back guarantee)Not ratedVisit
ElestioChecked
$11/moLinode 'Micro' (cheapest shown)1 vCPU · 1 GB RAM · 25 GB SSD3-day trial with $20 creditNot ratedVisit

n8n Cloud costs €20 a month for 2,500 workflow executions on the Starter plan and €50 a month for 10,000 on Pro, both billed yearly, as of November 1, 2026. Paying month to month costs €24 and €60. The self-hosted Community edition is free, so on your own server the only bill is the server itself, from about $6.49 a month.

How this guide was put together: plan prices, limits and the pricing FAQ come from n8n’s pricing page, and the rules for trials, credits and concurrency come from n8n’s documentation, all read on November 1, 2026. The annual prices on the page match AskGPU’s price data for n8n Cloud (checked October 17, 2026); the monthly-billing prices (€24, €60, €800) come from that data, since the page shows them only behind a toggle. Server prices come from AskGPU’s data for each host. We haven’t run n8n on these plans ourselves; the cost examples are arithmetic on the published numbers.

How much does n8n cost per month?

n8n sells three self-serve plans and one sales-led plan. All of them include unlimited users, unlimited workflows and every integration, and all are priced by how many workflow executions you run a month. n8n prices them in euros only.

Plan Billed yearly Billed monthly Executions a month Where it runs
Starter €20/mo €24/mo 2,500 n8n Cloud
Pro €50/mo €60/mo 10,000 n8n Cloud
Business €667/mo €800/mo 40,000 Self-hosted only (Cloud waitlist)
Enterprise Custom Custom Custom n8n Cloud or self-hosted
Community edition Free Free No cap listed Self-hosted

The pricing page’s toggle labels annual billing “Save 17%”. On Starter that’s 12 × €20 = €240 a year, against 12 × €24 = €288 paid monthly, a €48 difference. On Pro it’s €600 against €720.

There’s no USD price on n8n’s page. If your card is in dollars, your bank converts the euro charge at its own rate, so the dollar amount can move from month to month.

n8n’s Cloud docs also mention a “Pro-2” instance size with 50,000 executions. The public pricing page lists only the 10,000-execution Pro price, so check the plan picker in the Cloud console if you need more than 10,000 runs.

What counts as an execution?

This is the number that decides your plan. n8n’s pricing FAQ defines an execution as one run of an entire workflow, however many steps it has and however much data it moves. A workflow with 2 nodes and one with 20 nodes each count as one execution per run.

The FAQ gives rules of thumb for estimating:

  • A daily schedule: about 30 to 31 executions a month.
  • A schedule every 5 minutes: about 8,600 to 8,900 a month.
  • A webhook: one execution per incoming event, so count your daily events.
  • A chatbot: weekly conversations × messages per conversation, since each message triggers a run.

What n8n doesn’t publish on the pricing page is whether manual test runs, failed runs, sub-workflows and error workflows count toward the monthly quota. Its concurrency docs do say that manual, sub-workflow and error executions don’t count toward the concurrency limit, but that’s a separate limit (below). If your workflows call many sub-workflows, check the usage counter in the Insights dashboard during your trial.

This per-run counting is the main difference from Zapier, which counts a task for each action step. Our Zapier pricing guide explains its tiers.

What do Starter and Pro include?

Beyond the execution count, the plans differ on limits that matter once you have more than a handful of workflows.

Limit Starter Pro Enterprise
Concurrent executions 5 Up to 50 (table shows “20 / 50”) 200+
Shared projects 1 3 Unlimited
Assistant credits a month 1,600 Up to 9,600 Not included
Max execution duration 5 minutes 40 minutes 40 minutes
Saved executions 2,500 25,000 50,000
Execution log retention 7 days 30 days Unlimited
Workflow history 1 day 5 days 365+ days
Insights None 7 to 14 days Up to 365 days
Admin roles No Yes Yes
Support Forum Forum Dedicated, with SLA
Overage “Capped” “Capped” Custom

The page is inconsistent on Pro concurrency: the plan card says “up to 50 concurrent executions” and the feature table reads “20 / 50”. Starter also gets agents, MCP, AI nodes and “AI models without API keys” (Gateway credits, explained below). Pro adds admin roles, metric-based evaluations and workflow history.

n8n’s Cloud docs list the computing power behind each plan: 640 MiB of RAM on Starter and the 10,000-execution Pro, 1,280 MiB on the 50,000-execution Pro, and 4,096 MiB on Enterprise, each with up to 100 GB of storage. A workflow that handles large files can hit the memory ceiling on Starter before it hits the execution count.

What happens at the limit? When more production executions start than your concurrency limit allows, the extras wait in a queue and run in order as slots free up. When saved executions reach the count, storage or retention limit, n8n deletes the oldest ones; the FAQ says hitting those limits doesn’t stop workflows. For the monthly execution quota, the feature table says overage is “Capped” on Starter and Pro but doesn’t explain what that means in euros, and the FAQ only says “overage charges may apply if you do not upgrade to the next usage tier.” n8n doesn’t publish a per-execution overage rate.

What do Gateway and Assistant credits cost?

n8n Cloud has two credit balances, and they pay for different things.

Assistant credits Gateway credits
Pays for Chatting with n8n Assistant, including when it builds a workflow for you AI models and tool services that nodes call when your workflows run
Included 1,600 a month on Starter, up to 9,600 on Pro; resets on the 1st at midnight UTC A small one-time amount at sign-up (amount not published)
Unused credits Included credits don’t roll over Not applicable
Buying more Extra credit packs, used only after the monthly allowance Top-ups, manual or automatic
Expiry of bought credits 12 months after purchase 12 months after purchase
Plans Cloud Starter and Pro Cloud Starter and Pro

Gateway credits let a node call a supported model or service without your own API key. n8n bills per request at the rates on its service pricing page and says it aims to follow each provider’s published list price where one exists. n8n’s docs name model providers such as OpenAI, Anthropic and Google Gemini, and tool services such as Brave Search and Firecrawl (our report on the launch). When the balance hits zero, nodes using Gateway credits stop until you top up; nodes using your own credentials keep working. Top-ups are final and not refundable except where the law requires it.

Assistant credits are used per message based on the tokens the model processes, so long debugging sessions use more. Running a workflow never uses Assistant credits, even if the Assistant built it. If both included and extra credits run out, the Assistant pauses for everyone on the instance, but your workflows keep running.

Neither balance exists on self-hosted n8n or Cloud Enterprise. Self-hosted, you connect your own provider account and the provider bills you directly.

You can skip both: any AI node still accepts your own API key, and the model provider bills you at its own rates. For what those calls cost, see the AI token cost calculator.

Worked example: what four common workflows cost

Each example uses n8n’s own counting rules from the FAQ above.

1. A nightly report. One scheduled workflow, once a day: about 31 executions a month. Starter’s 2,500 covers it, with room for about 80 workflows like it (2,500 ÷ 31 ≈ 80). Cost: €20 a month billed yearly, or €240 a year.

2. A form-to-CRM workflow. 400 leads a month arrive by webhook, and each run has three action steps. That’s 400 executions on n8n, well inside Starter. The same workflow uses 3 tasks per lead on Zapier, so 1,200 tasks, which is above Zapier Professional’s 750-task entry tier ($19.99 a month billed yearly in our data).

3. A monitor that checks a site every 5 minutes. About 8,600 to 8,900 executions a month. That’s more than Starter’s 2,500, so it needs Pro: €50 a month billed yearly (€600 a year) or €60 monthly (€720 a year). Running it every 15 minutes instead cuts it to roughly a third, about 2,900 to 3,000, which still just misses Starter. Every 20 minutes (about 2,200) fits.

4. A support chatbot. 200 conversations a week × 6 messages each = 1,200 executions a week. A month averages about 4.33 weeks, so 1,200 × 4.33 ≈ 5,200 executions. That’s Pro. If the bot calls a model through Gateway credits, add the model’s per-request cost on top; with your own API key, the provider bills you instead.

Combine examples 1, 2 and 4 and you’re at about 31 + 400 + 5,200 = 5,631 executions, still inside Pro’s 10,000.

Is n8n free?

Yes, if you run it yourself. The Community edition is free to self-host with no execution cap listed, and n8n’s docs describe it as “almost the complete feature set”. Registering it with your email, also free, unlocks folders, debug in editor and custom execution data.

What the free edition leaves out: SSO (SAML, LDAP), projects, environments, Git version control, external secrets, log streaming, custom variables, multi-main mode and sharing. Without sharing, only the instance owner and the creator can open a workflow or credential. Queue mode and logging are included.

n8n Cloud has no free plan. The 14-day free trial gives Pro features with a cap of 1,000 executions, Starter’s computing power and a one-time Assistant allowance, and needs no credit card. If you don’t upgrade, the trial expires, nothing is charged and n8n deletes the instance; you have 90 days to download your workflows. A Business trial runs 14 days and does need a card.

The license. n8n uses its Sustainable Use License, not an open-source license. Per n8n’s license FAQ, you can use it free for your own business, build and run workflows for clients on your own instance, charge for setup and consulting, install it on a client’s own servers, and run it behind your product as long as only your backend talks to it. You can’t host n8n as a service where customers build their own workflows (through any UI, API, MCP or AI agent), fork or white-label it, or use license-key features without paying. Those uses need a commercial license.

What does self-hosting n8n cost?

The software is free; the server isn’t. From AskGPU’s price data:

Server Monthly Year one Notes
Hetzner CX23, 2 vCPU, 4 GB $6.49 $77.88 Excludes VAT and the IPv4 address; hourly billing, checked Oct 2026
Hostinger KVM 2, 2 vCPU, 8 GB $8.79 intro $105.48 Renews at $14.99/mo; prepaid term
Elestio on Netcup, 2 vCPU, 4 GB $16 $192 Managed: Elestio handles updates and backups
DigitalOcean 4 GiB Droplet, 2 vCPU $24 $288 Hourly billing, capped at the monthly price

Against Cloud Pro at €600 a year, a $77.88 Hetzner server is the cheapest route by a wide margin, with no execution quota. What you take on is the work: updates, backups, HTTPS and security. n8n’s 3.0 release, in beta since October 27, 2026, also drops npm installs, so self-hosting will need Docker (our report). Server sizing and a safe setup are in our self-hosting guide.

Self-hosted Business is the paid option for teams that need SSO, Git version control and environments on their own servers: €667 a month billed yearly (€8,004 a year) for 40,000 executions, or €800 monthly (€9,600 a year). It’s designed for companies under 100 employees. Companies under 20 employees can apply for n8n’s Startup plan, which the pricing page says gives 50% off Business. On the yearly price, half would be about €333.50 a month, but n8n’s page doesn’t state the billing terms of the discount.

Paid self-hosted licenses report usage: Business and Enterprise license keys ping n8n’s license server daily with execution counts. A license key works on unlimited instances, with combined usage counted against one quota.

How does n8n compare with Zapier and Make on price?

From AskGPU’s price data (n8n Cloud checked October 17, Zapier October 17, Make October 18, 2026):

n8n Cloud Zapier Make
Free plan None (self-hosting is free) 100 tasks/mo, two-step Zaps 1,000 credits/mo
Entry paid plan Starter €20/mo yearly Professional $19.99/mo yearly Core $12/mo yearly
Included at entry 2,500 executions 750 tasks 10,000 credits
What’s counted Each workflow run Each action step Credits used by modules
Users Unlimited 1 (Team plan for more) Teams plan for team roles

Because n8n counts runs and Zapier counts steps, the gap widens with longer workflows. Make sits between them, and its credit use depends on the modules in each scenario, so price it with your own workflow. Side-by-side tables: n8n Cloud vs Zapier and Make vs Zapier.

What to watch for

  • Prices are in euros. A US card pays a converted amount that can change each month, plus any foreign-transaction fee your bank charges.
  • No pro-rata refunds. The FAQ says monthly plans end at the close of the billing cycle and annual plans at the end of the term. Choosing yearly to save 17% locks you in for 12 months.
  • The trial deletes your instance. If you don’t upgrade within 14 days, n8n deletes the instance; download workflows within 90 days.
  • Overage terms are vague. “Capped” and “may apply” are all the page says. Watch your usage in Insights (Pro) as you approach the quota; Starter has no Insights dashboard.
  • Starter’s limits bite early. 5 concurrent executions, a 5-minute run time limit, 640 MiB of RAM and 1 day of workflow history.
  • Credits are final. Gateway and Assistant top-ups aren’t refundable, expire 12 months after purchase and are lost if you close the account.
  • Data location. n8n says Cloud data is stored in the EU (Frankfurt, Germany). If you need US data residency, self-host.
  • Self-hosted upgrades. Support for n8n 1.x ends December 31, 2026; plan the move to a current version.

Which n8n plan should you choose?

  • You can run a small Linux server: self-host the free Community edition on a $6.49 to $24 VPS. It’s the cheapest option at any volume, with no execution cap.
  • A solo user or small business with a few workflows: Cloud Starter, €20 a month billed yearly, if you stay under 2,500 runs and 5 at once. Pay monthly (€24) for the first months until your usage settles.
  • Frequent schedules, chatbots or a small team: Cloud Pro, €50 a month billed yearly for 10,000 runs, admin roles and Insights to track usage.
  • A company needing SSO and Git version control on its own servers: self-hosted Business at €667 a month billed yearly, or half that via the Startup plan if you have under 20 employees and qualify.
  • Light use that never needs code or self-hosting: compare against Zapier’s free plan and Make’s free 1,000 credits first. All options are on the automation hub, and the n8n plan data is on our n8n Cloud page.

Sources