If you’re deciding between n8n Cloud and self-hosting n8n yourself, the short answer is this: n8n Cloud costs €20 to €50 a month for a fixed number of monthly workflow executions and n8n handles the maintenance for you, while self-hosting the Community Edition is free software with no execution cap that puts server setup, updates, and uptime on you. The right pick comes down to three things: how much your time is worth, how much control you need over your data, and whether you’re already comfortable running a small server. Here’s how to work through that decision instead of guessing.
Builder · Level 2 of 3
n8n is a workflow automation platform: you connect apps and services together in a visual, node-based builder, similar in spirit to Zapier or Make, but built to also run AI agents with memory and tool access. You can use n8n two ways — as a hosted product n8n runs for you (“n8n Cloud”), or as free, self-hosted software you install on your own server. Both run the identical editor and node library. The choice only changes who’s responsible for keeping it running, and what it costs you each month.
What n8n’s AI agent tooling looks like now
This decision matters more than it used to because n8n isn’t just a task-connector anymore. Its AI Agent node pairs a language model with memory and a library of prebuilt tools, and n8n’s node library has grown past 500 integrations, including support for MCP servers so an agent can call other agents or external tool servers. You get human-in-the-loop approval steps for anything you don’t want the agent doing unsupervised, and every Cloud plan now bundles a separate in-editor AI Assistant (currently in preview) that helps you build workflows, with its own monthly credit allowance layered on top of your execution limit.
That AI Assistant is Cloud-only for now. Self-hosted instances don’t get it yet, so if you self-host and want AI help while building, you’re on your own tools and your own API key. It’s a real point in Cloud’s favor if in-editor AI assistance matters to you, and a non-issue if it doesn’t.
Real cost comparison: cloud plans vs VPS and time
n8n bills Cloud plans on workflow executions, not on individual steps — one run of an entire workflow counts as one execution no matter how many nodes it touches. That makes usage more predictable than per-task pricing, but it still caps out:
| Plan | Monthly cost | Executions included | What you get |
|---|---|---|---|
| Cloud Starter | €20/mo, billed annually | 2,500 | 1 shared project, 5 concurrent executions, 2,300 AI Assistant credits, forum support |
| Cloud Pro | €50/mo, billed annually | 10,000 | 3 shared projects, 20 concurrent executions, admin roles, workflow history |
| Self-hosted Community Edition | Free (plus your server cost) | Unlimited | Every core feature, no AI Assistant yet, you run and maintain it |
Pricing and plan details above were checked directly against n8n’s official pricing page on 2026-07-16.
The other number worth understanding is concurrent executions: how many workflows n8n will run for you at the exact same moment. Starter allows 5 at once, Pro allows 20. For most personal automations (a daily digest, a form-to-Slack notifier, a handful of scheduled scrapes) you’ll rarely brush against 5, let alone 20. It only becomes a real constraint if you’re triggering a lot of webhook-based workflows in quick bursts — for example, processing every order from a busy storefront the moment it comes in. If that’s your use case, run the numbers on Pro or self-hosted before you build, not after you hit the ceiling mid-launch.
Self-hosting removes the execution ceiling, but it isn’t actually free — you’re trading a subscription for a small VPS bill plus your own setup time. Budget VPS providers commonly price an entry-level box big enough to run n8n comfortably somewhere in the $5 to $12 a month range, well under either Cloud plan. The real cost is the afternoon it takes to provision the server, install n8n, and get HTTPS working, plus the recurring time you’ll spend on updates afterward. If you want the exact walkthrough and current pricing for that route, the self-hosting on a budget VPS guide covers it step by step.
What determines where you land in that VPS price range is mostly RAM and how many workflows run at once. A small droplet with 1 to 2 GB of memory handles a handful of scheduled automations without trouble, while running heavier AI-agent workflows with larger memory buffers pushes you toward a bigger, slightly pricier box. Either way, you’re choosing the server size once and then living with it, versus a Cloud plan you can upgrade or downgrade in a couple of clicks.
Maintenance and uptime tradeoffs
On Cloud, n8n handles version upgrades, scaling, and uptime. If a workflow fails because of a platform issue, that’s their problem to fix, not yours. On a self-hosted box, all of that maintenance is yours: applying n8n updates, patching the underlying OS, watching disk space, and noticing when the service quietly stops responding at 2 a.m. None of it is difficult work individually, but it’s ongoing, and it’s easy to let it slide until something breaks mid-automation.
A reasonable middle ground: self-host, but set up basic uptime monitoring and automatic backups from day one, so a dead server is a quick restore instead of a rebuild. If that sentence sounds like more setup than you want, that’s a genuine signal to stay on Cloud.
Data control and privacy considerations
n8n Cloud stores your workflow and execution data on servers in Frankfurt, Germany, inside the EU. For most personal and small-business automations, that’s a non-issue. It matters more if you’re routing anything sensitive through your workflows: client data under a contract that names a specific storage region, health or financial information, or anything you’d rather never leave hardware you control. Self-hosting puts your data wherever you put your server, full stop, which is the whole appeal for privacy-conscious builders and the reason a lot of “self-hosted AI agent” setups exist in the first place.
Either way, treat your credentials seriously. n8n encrypts stored secrets on both Cloud and self-hosted instances, but on a self-hosted box you’re also responsible for securing the server itself: firewall rules, SSH access, and keeping the host OS patched.
You’re not locked into your first choice
Workflows built in n8n export as plain JSON, and the editor is identical whether it’s rendered from Cloud or from a self-hosted instance. That means starting on Cloud and moving to self-hosted later, or the reverse, is a real export-and-import job, not a rebuild from scratch. If you’re not sure which side you’ll land on long-term, that portability is your safety net: pick the lower-friction option now, and let actual usage tell you whether it’s worth switching in three or six months. The one thing that doesn’t carry over automatically is your execution history and any credentials you’ve stored, so budget an hour to reconnect accounts and re-enter API keys on the new instance.
There’s also a hybrid option worth knowing about: n8n’s Business plan supports separate dev, staging, and production environments, which is how some builders run a rough draft on Cloud, then promote the finished, stable workflow to a self-hosted production instance once it’s proven itself. That’s more setup than most personal-automation readers need, but it’s there if you’re supporting a small team.
A simple decision framework by reader type
Match yourself to the closest description:
- You’re testing an idea and don’t want ops work: start on Cloud Starter. €20 a month buys you zero maintenance while you find out if the automation is worth keeping.
- You already run a home server, NAS, or VPS for other projects: self-host. The marginal effort is low if you’re already comfortable in that world, and you keep every execution free.
- You need data to stay in-house for compliance, client, or personal-privacy reasons: self-host. This is close to a hard requirement, not a preference.
- You run enough workflows to blow past 10,000 executions a month and don’t want to manage a server: compare Cloud Pro’s overage terms against a self-hosted setup on a slightly bigger VPS — at that volume, the math is worth doing by hand.
- You want the in-editor AI Assistant while you build: that’s Cloud-only right now, which tips the decision if it matters to your workflow.
Most people land in the first two categories. If you’re still unsure, the honest default is to start on Cloud’s free trial, build the workflow you actually need, and only move to self-hosting once you know it’s a keeper and you understand what it costs in executions each month.
Where this fits with Zapier and Make
This guide assumes you’ve already decided n8n is the platform, not which no-code tool to use in the first place. If you’re still weighing n8n against the alternatives, the Zapier vs Make vs n8n comparison covers pricing model, learning curve, and where each one actually wins before you commit to n8n’s cloud-vs-self-host question at all.
Either path gets you to the same editor and the same node library. Pick based on your time, your data rules, and how much you already enjoy running a small server, not on which option sounds more impressive.