n8n Pricing Explained: What It Actually Costs a Solopreneur to Run (2026 Honest Breakdown)

n8n pricing confuses more solopreneurs than any other line item in the automation world — and it’s not because the numbers are high. It’s because the pricing model is nothing like Zapier’s, the “free” version is genuinely free but hides a cost most people don’t see coming, and almost every article ranking for this term is either the official pricing page or a vendor trying to sell you hosting. So let me do what none of them do: give you the real numbers, decode the one concept everyone trips on, and show you my actual bill for running n8n across a 10-brand automation fleet.

I’ve been self-hosting n8n in production for months — it’s the connective tissue between my Claude agents and the outside world. I pay for exactly none of the Cloud plans, and I’ll explain precisely when that’s the right call and when it absolutely isn’t. No affiliate links, no upsell. Just what it costs and what you get.

n8n Pricing at a Glance: The Plans, Decoded

Here’s the whole menu in plain English. n8n sells four hosted tiers plus a free self-hosted edition. All prices below are the billed-annually rate (paying monthly costs roughly 17% more), current as of 2026:

• Community Edition — Free. The full open-source n8n, self-hosted on your own server. Unlimited executions, unlimited workflows, every integration. The catch isn’t money — it’s maintenance (more on that below).
• Starter — $20/mo. 2,500 workflow executions per month, 5 concurrent executions, 1 shared project, 2,300 AI credits. Hosted by n8n, so zero setup.
• Pro — $50/mo. 10,000 executions, 20 concurrent, 3 shared projects, 7 days of insights, up to 13,700 AI credits, admin roles, workflow history, and execution search. This is the “solo builder in production” tier.
• Business — $800/mo. 40,000 executions, 6 projects, SSO/SAML/LDAP, Git-based version control, and different environments. This is where n8n jumps from “solopreneur” to “company with a compliance department.”
• Enterprise — Custom. Unlimited-ish scale, 200+ concurrent executions, 365 days of insights, dedicated support with an SLA. If you’re asking the price, it’s not for you yet.

Two things jump out immediately. First, the leap from Pro ($50) to Business ($800) is a 16x cliff with almost nothing in between — that gap is the single most important fact in this whole article, and it’s exactly why self-hosting matters for people like us. Second, notice what you’re paying for: executions, not tasks, not steps, not users. That distinction is where everyone gets burned, so let’s kill the confusion right now.

A word on the free trial and AI credits

Every Cloud plan starts with a free trial and no credit card required, so you can pressure-test your real execution volume before you pay a cent — do that. The “AI credits” bundled into Starter (2,300/mo) and Pro (up to 13,700/mo) are for n8n’s built-in AI nodes; they’re a convenience, not a reason to pick a tier. If you’re calling your own OpenAI or Anthropic keys inside workflows — which most serious builders do — those credits are largely irrelevant and you’re billed by the model provider directly, not by n8n. Don’t let a big AI-credit number talk you into a plan your execution count doesn’t justify.

The One Thing That Trips Everyone Up: Executions, Not Tasks

If you’re coming from Zapier, your brain is wired to think in tasks — every single step of every automation costs you one billing unit. Send an email? One task. Update a spreadsheet row? Another task. A five-step Zap that runs once burns five tasks. That model punishes complexity, so people build tiny, timid automations.

n8n throws that model out. You pay per execution — one complete run of a workflow from trigger to finish, no matter how many steps are inside it. A workflow with 3 nodes and a workflow with 50 nodes both cost you exactly one execution when they run. This is n8n’s actual competitive advantage, and it changes how you build: you stop fragmenting your logic to save on task counts and start building fat, capable workflows that do real work in a single run.

But here’s the trap on the other side. Because executions are cheap conceptually, people forget how often their workflows fire. Do the math that nobody does: a single workflow triggered every 5 minutes runs 12 times an hour, 288 times a day, roughly 8,640 times a month. That one workflow alone blows through the Starter plan’s 2,500 executions in about nine days. Add a couple of webhook-triggered flows that fire on every form submission or every new email, and Pro’s 10,000 executions evaporate faster than you’d guess.

So the real question isn’t “how many workflows do I have?” It’s “how many times per month will they all run, combined?” To make that concrete, here’s roughly what common trigger patterns cost you per month, per workflow:

• Every 1 minute: ~43,200 executions/mo — blows past every plan except Business. Almost always a mistake; poll less often.
• Every 5 minutes: ~8,640 executions/mo — one workflow nearly maxes out Pro on its own.
• Every 15 minutes: ~2,880 executions/mo — a single flow already exceeds Starter.
• Hourly: ~720 executions/mo — three or four of these fit inside Starter comfortably.
• Once a day: ~30 executions/mo — you could run dozens of these and never feel it.
• Per-webhook (form submits, new emails): entirely dependent on your traffic — a busy contact form or inbox can quietly out-execute a scheduled job.

Add up your real triggers using that table and you’ll know your plan in about two minutes. If the total is comfortably under 2,500, Starter is fine. If you’re scheduling frequent polling jobs or high-traffic webhooks, you’ll be shopping for a bigger plan fast — or looking at the option that removes execution limits entirely.

Self-Hosting n8n: What It Really Costs (The $5 VPS Math)

This is the part the official pricing page will never tell you: the Community Edition is free, self-hostable, and has zero execution limits. You can run a million executions a month and pay nothing to n8n. The only cost is somewhere to run it. Here are the realistic options, cheapest first:

• Managed one-click hosting — $3-7/mo. Platforms like PikaPods (~$3.80/mo) and InstaPods ($3-7/mo) deploy n8n with a single click — HTTPS, updates, and backups handled for you. Unlimited executions, near-zero maintenance. For most solopreneurs, this is the sweet spot: about one-eighth the cost of n8n Cloud Starter with none of the execution ceiling.
• Raw VPS + Docker — $4-5/mo. Rent a small box (a Hetzner CX22 runs about $4.51/mo), install Docker, and run n8n with Docker Compose. Total control, unlimited executions. Setup is 30-60 minutes if you’re comfortable with Docker — and this is exactly how I run it.
• Coolify on a VPS — $5-8/mo. Coolify gives you a Heroku-like control panel on your own server: GUI deployments, automatic SSL, basic monitoring. A nice middle ground between one-click and raw sysadmin work.

So the honest self-hosted number is roughly $5 a month, all in. But “free software on a $5 box” hides one real cost: your time. When you self-host raw, you’re the sysadmin. OS updates, Docker updates, n8n version upgrades, SSL renewals, backups — that’s your job now. When n8n ships a breaking change or your certificate expires at 2 a.m., nobody’s coming to fix it. Managed one-click hosting exists precisely to buy that headache away for a couple of extra dollars, which is why I usually point beginners there first. If you want the deeper mechanics of wiring n8n into an agent stack, I broke that down in my guide to building a no-code AI agent.

n8n Cloud vs Self-Hosted: The Honest Decision Framework

Forget the fanboy takes. Here’s the actual decision, reduced to the questions that matter:

Pay for n8n Cloud if: you don’t want to think about servers at all, your execution volume fits comfortably inside Starter or Pro, you want official support and the managed AI credits, or you’re in a business where “who patches the server” is a real liability question. Cloud is a completely legitimate choice — you’re buying time and peace of mind, and $20-50/mo is nothing against the hours you’d spend maintaining infrastructure.

Self-host if: your execution volume is high or unpredictable, you’re comfortable with (or willing to learn) Docker, you want unlimited runs for the price of a coffee, or you’re building something you intend to scale. The moment your workflows would push you toward that $800 Business tier, self-hosting isn’t just cheaper — it’s the only sane option. That Pro-to-Business cliff is where self-hosting stops being optional.

My rule of thumb: if n8n is a nice-to-have, pay for Cloud. If n8n is load-bearing infrastructure for your business, own it. For me it’s load-bearing — it’s how my agents talk to webhooks, APIs, and each other — so I self-host. If you’d rather skip the setup entirely and have someone wire the whole thing up for you — n8n, the agents, the works — that’s literally what I do for clients. You can book an automation strategy session and we’ll map it out. Either way, get the architecture right before you worry about the bill; I covered the broader AI-first approach to workflow automation if you want the bigger picture.

My Actual n8n Bill Running a 10-Brand Fleet

Here’s the receipt, because talk is cheap and I’d rather show you. I run automations across a fleet of autonomous brand containers — this brand is one of ten. n8n is the glue: it catches webhooks, pings APIs, wakes up Claude agents on a schedule, and passes data between systems that were never designed to talk to each other.

My total n8n spend for all of it? The cost of the VPS it runs on — about $5 a month. One self-hosted instance, Docker Compose, unlimited executions, serving every brand. If I’d put that same workload on n8n Cloud, the execution volume alone would have shoved me straight past Pro and into pricing-department territory. Instead it’s a rounding error, and I’ve written before about how the real cost of running AI agents is almost always the model calls, not the plumbing.

In a given month my n8n instance handles tens of thousands of executions across the fleet — scheduled agent wake-ups, form and email webhooks, cross-brand data hand-offs. On Cloud, that volume would have meant the Business tier or an overage conversation. Self-hosted, the number is invisible; there is no counter to watch.

The trade-off is real and I won’t pretend otherwise: I own the maintenance. When n8n releases an update, I’m the one who pulls the new image and checks nothing broke. In practice that’s maybe fifteen minutes a month — I pin the version, read the changelog before upgrading, and keep a nightly backup of the workflows and credentials so a bad update is a five-minute rollback, not a crisis. That’s a fair deal at my scale because the automations are core to how the business runs — the same reason I keep a webhook that wakes my Claude agents automatically. If n8n were a weekend experiment instead of production infrastructure, I’d happily pay someone $5-7/mo to manage it for me and never think about it again. Match the hosting choice to how much the automation actually matters to your income.

n8n Pricing vs Zapier vs Make: The Real Cost-Per-Workflow Comparison

The only comparison that matters is cost-per-actual-work, not sticker price. Here’s how the three stack up for a solopreneur:

• Zapier charges per task (per step). It’s the easiest to start with and the fastest to get expensive — complex, multi-step automations that run often will run up your bill quickly, because every step counts. Great for simple, low-volume connections; punishing at scale.
• Make charges per operation (roughly per module run), which is more granular than Zapier’s per-task model and often cheaper for the same automation. Its visual builder is genuinely nice. But you’re still metered on internal complexity, so heavy workflows still cost more as they grow.
• n8n charges per execution on Cloud (complexity is free) — or nothing but hosting if you self-host. For anyone running frequent or complex automations, this is structurally the cheapest model, full stop. The more sophisticated your workflows get, the more n8n’s execution-based pricing pulls ahead.

The short version: for occasional, simple automations, any of the three is fine and Zapier is the gentlest on-ramp. But the moment your automation becomes a real system — many steps, running many times a day — n8n’s model wins on cost, and self-hosted n8n wins by a landslide. If you’re wiring n8n into a larger agent setup, my n8n + MCP guide for Claude agents shows how the pieces connect, and it’s part of the broader actual stack I use to run my brands.

FAQ: n8n Pricing, Answered Straight

Is n8n free forever?

Yes — the Community Edition is genuinely free forever, self-hosted, with unlimited executions and workflows. You only pay if you choose n8n Cloud (managed hosting) or need enterprise features like SSO. “Free” means free software; you still need a ~$5/mo server to run it.

Do n8n executions roll over?

No. Cloud execution allotments reset each billing period and unused executions don’t carry forward. If you consistently run under your limit, you’re on too big a plan; if you’re hitting the ceiling, upgrade or self-host to remove the limit entirely.

What counts as one execution?

One complete run of a single workflow, from trigger to finish — regardless of how many nodes or steps are inside it. A 50-node workflow and a 3-node workflow each cost one execution per run. Frequency of runs, not complexity, drives your usage.

When should I upgrade from Starter to Pro?

When you consistently exceed ~2,500 executions a month, need more than one project, or want workflow history and execution search for debugging. If you’re blowing past Pro’s 10,000 executions too, that’s your signal to self-host rather than climb toward the $800 Business tier.

Is self-hosting n8n hard?

Managed one-click hosting (PikaPods, InstaPods) is genuinely easy — a few clicks and you’re live. Raw VPS + Docker takes 30-60 minutes and some comfort with the command line. The ongoing work is updates and backups; one-click hosts handle that for you, which is why they’re my recommendation for anyone who doesn’t want to be a part-time sysadmin.

Final Thoughts: Pay for Time or Pay for Control

Strip away the confusion and n8n pricing comes down to one honest choice: pay n8n for convenience, or pay ~$5 for a server and own your automation outright. There’s no wrong answer — there’s only the answer that matches how much this stuff matters to your business. If it’s a helpful side tool, grab Cloud Starter and get on with your day. If it’s the backbone of how you operate, self-host it and never look at an execution counter again.

What you should not do is guess. Estimate your monthly execution volume, pick the model that fits, and revisit it when your usage changes. That five minutes of math is the difference between a $5 bill and an $800 surprise. And if you’d rather I just build the whole automation layer for you — n8n, the agents, the pipelines that run while you sleep — book an automation strategy session and let’s get it wired properly.