n8n is an open-source workflow automation platform whose self-hosted Community Edition is free forever with unlimited executions, workflows, and users — you pay only for a server, typically $4-7/month. Cloud starts around $24/month for 2,500 executions. The critical detail: when you hit a Cloud execution cap, workflows stop entirely until the next billing cycle. There's no overage option and no grace period.
Reviewed by EdGrows·Updated Aug 2, 2026Affiliate link — we may earn a commission
This analysis is based on documentation, public user reports, and vendor materials — not yet on our own hands-on testing. How we rate
n8n is open-source workflow automation — connect apps, move data, run logic, trigger AI. Think Zapier, except you can download the whole thing and run it on your own server for free.
That last part isn't a marketing framing. The Community Edition is genuinely free: unlimited executions, unlimited workflows, unlimited users, full core features. You pay for a server, which runs $4-7 a month on a VPS.
There's also a managed Cloud, starting around $24 a month for 2,500 executions.
The gap between those two options is bigger than the price difference suggests, and it comes down to one mechanic worth understanding before you sign up for anything.
Most metered services handle overruns gracefully. You get a warning email, then throttling, or an overage charge, or a grace period.
n8n Cloud does none of that. Hit your monthly execution cap and every workflow stops. No warning. No grace period. No option to pay for more. Everything sits still until the billing cycle resets.
For a hobby project that's annoying. For automations running actual operations — lead routing, order processing, alerting — it's a genuine outage you didn't schedule.
And the caps arrive faster than they look, because triggers fire on schedules rather than when you're paying attention:
| Trigger pattern | Monthly executions |
|---|---|
| Every 5 minutes | 8,640 |
| Every 15 minutes | 2,880 |
| Gmail polling (typical) | 2,500 in under 11 days |
| Five modest workflows | ~18,750 |
Read that last row against Starter's 2,500 and Pro's 10,000. Five ordinary automations put you past both — you'd need Business at roughly €667 a month to run them on Cloud without hitting the wall.
The same five workflows run without limits on a $5 VPS.
Having led with the warning, credit where it's due — because n8n's billing unit is the fairest in the category.
One execution equals one workflow run, regardless of steps. A ten-step workflow triggered once costs one execution.
Zapier counts every action as a task. That same ten-step workflow costs ten. Make counts operations similarly.
So when you compare prices, you can't compare the headline numbers directly. Multiply your Zapier task count by your average workflow length to get the n8n-equivalent figure. Do that honestly and n8n almost always wins on per-unit cost for anything multi-step — often dramatically. Teams hitting 10,000 runs a month on n8n Pro would be looking at $300+ for the rough equivalent on Zapier Professional.
There's a second lever most people miss: batching. Instead of triggering on every new lead at 100 executions a day, process all new leads at 9am and 5pm — two executions. That's a ~98% reduction for anything batch-friendly, and it's often the difference between needing Pro and staying on Starter.
n8n fits technical teams who can run a Docker container. If that describes you, self-hosting gives unlimited automation for the cost of a coffee, and the ceiling disappears entirely. Developers, technical founders, and ops people with any infrastructure comfort get the best deal in this category by a wide margin.
It also suits teams who need data residency or control — self-hosting means workflows and data never leave infrastructure you own, and Cloud runs on EU servers in Frankfurt for GDPR purposes.
And AI workflow builders specifically: n8n has become a common backbone for agent and LLM pipelines, where its code nodes and open architecture beat closed rules-based tools.
Where it's the wrong call: non-technical users who want everything managed with no surprises (Zapier is easier and more forgiving), teams needing predictable costs on Cloud without infrastructure work, and anyone whose automations are simple enough that Zapier's polish is worth the premium.
| Tool | Billing unit | Self-host | Free option | Entry paid | Overage behavior |
|---|---|---|---|---|---|
| n8n | Workflow run | ✅ Free | ✅ Self-hosted | ~$24/mo | ❌ Hard stop |
| Zapier | Per task/action | ❌ | ✅ Limited | ~$20/mo | ⚠️ Throttle/upgrade |
| Make | Per operation | ❌ | ✅ Limited | ~$9/mo | ⚠️ Buy more |
| Lindy | Credits | ❌ | ❌ Trial only | $49.99/mo | ⚠️ Usage-based |
| Botpress | Conversations | ⚠️ Some | ✅ Yes | ~$89/mo | ⚠️ Auto top-up |
Pricing checked August 2026 from each provider's pages. n8n prices in euros, so dollar figures vary with exchange rates. Competitor prices are approximate.
n8n vs Zapier: Zapier is easier, more polished, and has more integrations — genuinely better if you're non-technical and your automations are simple. n8n is dramatically cheaper for multi-step work because of the execution model, and free entirely if you self-host. The honest split: Zapier for people who want it handled, n8n for people who want it cheap and controllable.
n8n vs Make: Make has a lower entry price and a strong visual builder, but bills per operation like Zapier, so multi-step workflows cost more than they appear to. n8n's code nodes and self-hosting go further for technical users. For visual complexity without infrastructure, Make; for cost and control, n8n.
n8n vs Lindy: Lindy builds AI agents from natural language for non-developers, with no free tier and credit-based costs. n8n is a workflow engine you assemble deliberately. For agents without technical work, Lindy; for automation you architect yourself, n8n.
n8n vs Botpress: Botpress is conversation-focused with its own agent platform. n8n is general-purpose automation that happens to handle AI well. Different jobs — plenty of teams run n8n as the workflow layer behind a conversational tool.
| Option | Cost | Executions | Notes |
|---|---|---|---|
| Self-hosted Community | Free + $4-7/mo server | ♾️ Unlimited | Unlimited workflows and users too |
| Cloud Starter | ~€24/$24/mo | 2,500 | ~83/day; 5 concurrent workflows |
| Cloud Pro | ~€60/$60/mo | 10,000 | Small teams and agencies |
| Business (self-hosted) | ~€667-800/mo | 40,000 | SSO, Git, multi-environment |
| Startup (Business, 50% off) | ~€333/mo annual | 40,000 | Under 20 employees, under €5M funding |
| Enterprise | Custom | Unlimited | 200+ concurrent, SLA, log streaming |
Pricing checked August 2026. n8n prices in euros; dollar equivalents shift with exchange rates, which is why sources quote $20 and $24 for the same plan. Annual billing saves roughly 17%. No permanent free tier on Cloud — 14-day trial only. Extra execution buckets on Business run about €4,000 for 300,000.
The decision is genuinely simple once you know two things.
If you can run Docker, self-host. Roughly $5 a month on a VPS, unlimited everything, no caps to monitor. That's about an eighth of Cloud Starter's price with none of the constraints, and it saves most teams $200-700 a year. Managed one-click hosts exist if you want the middle ground.
If you can't, size Cloud with real headroom and use batching aggressively. Count your triggers, do the arithmetic above, and pick a tier with margin — because there's no overage safety net, only a stop.
Two things worth knowing beyond that. The 2026 change making workflows and users unlimited on every Cloud plan is genuinely good — your bill now depends on one number instead of three, and adding teammates is free. And the startup discount at 50% off Business is the cheapest legitimate path to self-hosted SSO and Git version control if you qualify.
One transparency note: several of the loudest voices comparing n8n Cloud unfavourably to self-hosting sell managed n8n hosting. Their cost arithmetic holds up under checking — but their conclusions are also their business model.
AI agent pipelines: A team builds LLM workflows with vector stores and model calls as native nodes, using code steps where visual logic runs out.
Lead routing at volume: Inbound leads get enriched, scored, and routed — batched to twice daily rather than triggered per lead, cutting executions by roughly 98%.
Self-hosted data workflows: A company with residency requirements runs everything on its own infrastructure, so customer data never touches a third-party automation vendor.
Replacing an expensive Zapier bill: A team doing multi-step automation migrates and pays a fraction, because ten-step workflows count once rather than ten times.
Internal ops glue: Engineering connects internal tools with HTTP nodes and custom code where no prebuilt integration exists.
n8n is the best value in workflow automation, and it isn't close for anyone technical. The Community Edition is genuinely free with no execution, workflow, or user limits — run it on a $5 VPS and you have unlimited automation with full control over your data. The execution model, counting a workflow run once rather than per step, is structurally fairer than Zapier's or Make's and saves real money on anything multi-step. Code nodes and native AI tooling make it the default backbone for agent pipelines in 2026.
The Cloud side needs going into with eyes open. Hitting your execution cap stops every workflow with no warning, no grace period, and no overage option — harsher than any major competitor, and the most common complaint from new users. Caps arrive quickly, since a single five-minute trigger consumes 8,640 executions monthly. There's no permanent free tier on Cloud anymore, only a 14-day trial. And the platform genuinely rewards technical comfort; non-technical users will find Zapier easier and more forgiving.
For anyone who can run a container, self-host and stop thinking about it. For everyone else, Cloud works — size it with headroom, batch your triggers, and watch the dashboard.
Note: n8n is open source and does not have an affiliate program with AIVario. We earn no commission, and this rating carries no commercial incentive.
Best for: Technical teams who can run Docker, anyone building AI agent pipelines, teams needing data residency or self-hosting, high-volume multi-step automation, startups qualifying for the 50% Business discount Not ideal for: Non-technical users wanting managed simplicity, teams needing overage safety on Cloud, simple automations where Zapier's polish justifies the premium Bottom line: Free and unlimited if you can run a container, which makes it the best value in automation. On Cloud, the hard stop at your cap is the thing to plan around — size up and batch your triggers.
The self-hosted Community Edition is genuinely free — open source, no execution limits, no workflow limits, no user limits, and full access to core features. What isn't free is running it: you need a server, which costs roughly $4-7/month on a VPS, or around $3-7 on a managed one-click host. That's the honest total. For technical teams, it's the lowest cost of ownership in workflow automation by a wide margin, typically saving $200-700 a year versus Cloud.
Starter runs around €24/$24 per month for 2,500 executions and Pro around €60/$60 for 10,000, with a self-hosted Business plan near €667-800 for 40,000 executions and custom Enterprise pricing. Annual billing saves roughly 17%. Note that n8n prices in euros, so US figures shift with the exchange rate — which is why you'll see $20 and $24 quoted for the same plan across different sources.
Everything stops. n8n Cloud halts all your workflows the moment you exceed the monthly cap — no warning, no grace period, and no option to pay an overage. Automations simply don't run until the next billing cycle resets. This is the most common shock new Cloud users report, and it's meaningfully harsher than competitors who throttle or bill overages. If your automations matter operationally, size your plan with genuine headroom or self-host.
Faster than the numbers suggest, because triggers fire on schedules rather than on your attention. A single workflow polling every five minutes consumes 8,640 executions a month — over three times the Starter allowance from one automation. A Gmail polling workflow can exhaust Starter's 2,500 in under eleven days. Five modest workflows can reach 18,750 monthly executions, which is 7.5 times the Starter cap and nearly twice Pro's. Count your triggers before picking a tier.
No, and this is n8n's structural cost advantage. n8n counts one workflow run as one execution regardless of how many steps it contains — a ten-step workflow triggered once is one execution. Zapier counts each action as a task, so the same workflow would consume ten. When comparing prices, multiply your Zapier task count by your average step count to get the equivalent n8n figure. For multi-step automation, n8n wins on per-unit cost almost every time.
Yes, and batching is the single biggest lever. Instead of triggering on every new lead — say 100 executions a day — process all new leads at 9am and 5pm, which is two executions daily. That's roughly a 98% reduction for anything batch-friendly. Combining multiple triggers into scheduled runs, and checking n8n's per-workflow execution dashboard weekly, will usually keep a team on a lower tier than their raw trigger count suggests.
If you can run a Docker container, self-host. You get unlimited executions for the price of a VPS, which is roughly an eighth of Cloud Starter with none of the caps. Cloud makes sense in three situations: you genuinely don't want to touch infrastructure, you're running under 2,500 executions monthly, or you need managed hosting for compliance reasons. Note that many of the loudest self-hosting advocates online sell managed n8n hosting, so weigh their conclusions accordingly — though the underlying cost math holds up.
The most user-friendly change in a while: every Cloud plan now includes unlimited workflows and unlimited users. Previously those were tiered. Now the only lever on your bill is execution volume, which makes plan selection much simpler — you're forecasting one number instead of three. It also means adding teammates costs nothing, unusual in a category where per-seat pricing is the norm.
Yes, and it's substantial if you qualify. Companies with fewer than 20 employees and under €5M in total funding can get 50% off the self-hosted Business plan — roughly €333/month billed annually instead of €667. That unlocks 40,000 executions, shared projects, SSO/SAML/LDAP, and Git version control via a license key, valid for a year and renewable while you still qualify. It's the cheapest legitimate route to self-hosted SSO and version control.
Anthropic's AI for deep reasoning, long documents, and writing that doesn't sound like AI.
AI meeting notes that combine your rough typing with audio transcript into structured summaries. The smart note-taking tool that won 2025.
AI meeting recorder for Zoom, Google Meet, and Teams. Unlimited free recording; AI summaries capped at 5/month on the free tier.