n8n Pricing in 2026: A Plain-English Breakdown

Last updated: April 2026

If you've landed here, one of two things is probably true. Either you saw n8n's marketing — "fair-code, self-hostable, free forever" — got interested, opened the pricing page, and felt mildly confused by the tier matrix. Or you've been using n8n Cloud for a few months, your execution count is creeping toward your cap, and you're trying to figure out what the jump to the next tier actually costs.

Both are reasonable places to arrive. n8n's pricing is more honest than most of the iPaaS category — it doesn't price per step the way Zapier does — but it's not simple. There are four public tiers, an enterprise option, and a self-hosted Community edition that's technically free but quietly expensive. This guide walks through each tier with current 2026 numbers, the hidden costs most blog posts skip, who n8n is genuinely a good fit for, and who it isn't.

No hit piece here. n8n is a legitimately good product with real strengths. The goal is just to help you decide whether it's the right tool for your situation before you commit six months of workflow-building to it.

n8n's Pricing Tiers (2026)

Here's the full public pricing, current as of April 2026. Annual billing saves roughly 17% over monthly; the numbers below are the annual prices except where noted.

A few things worth highlighting that the marketing page doesn't put front and center:

Executions are per workflow run, not per step. This is the single most important thing about n8n's pricing model. A 50-node workflow that triggers once counts as one execution. The same workflow in Zapier would burn 50 tasks. This is why teams running complex workflows almost always save money switching from Zapier to n8n, even on Cloud.

Sub-workflows count separately. If your main workflow calls a sub-workflow, that's two executions. n8n's "Execute Workflow" node is convenient for reusing logic, but every call bills.

Failed runs still count. If a workflow fails on step two of forty, you've still used an execution. Retries also count.

Business tier pricing has two quoted numbers. €667/month when billed annually (€8,000/year upfront). €800/month if you pay monthly. Both are listed on the pricing page; skim too fast and you'll miss the annual discount.

Enterprise is a quote. If you need SOC 2 Type II attestation, HIPAA BAAs, dedicated support, or on-prem deployment with commercial license, you're talking to sales. Expect pricing in the €2,000–€6,000/month range depending on volume and features.

What the Community (self-hosted) tier actually gives you

The self-hosted edition is n8n's biggest structural advantage over Zapier and the main reason technical teams pick it. You get the full product — every node, every integration, full workflow builder — running in your own Docker container, VPS, or Kubernetes cluster. No execution limits other than what your hardware can handle.

What you don't get: SSO, SAML, LDAP, audit logs, external secrets, log streaming, or commercial support. Those sit behind the Enterprise license. For most internal-use cases — a five-person ops team running 30 workflows — you don't need any of that. For a 500-person company with a security team, you do.

The Hidden Costs of n8n

The tier table above is the sticker price. The actual total cost of ownership usually runs higher. This section is the part that Twitter threads about "just self-host n8n, it's free" tend to skip.

Self-hosting infrastructure

If you run n8n Community on your own infrastructure, you need at minimum:

• A server. n8n runs on Node.js. A small workload fits on a $5–$10/month Hetzner or DigitalOcean VPS. A mid-sized workload — say, 20,000 executions/month with a few memory-hungry workflows — wants 4 GB RAM and 2 vCPUs, which runs $20–$40/month. High-volume self-hosting (100k+ executions) usually needs queue mode with a separate worker instance, Redis, and a dedicated Postgres instance; figure $100–$300/month for infrastructure alone.
• A database. SQLite is fine for tiny workloads but every production deployment should use Postgres. Managed Postgres is $15–$50/month at the entry level.
• Backups. If the server dies, you've lost your workflow definitions and execution history. S3 or equivalent, plus a backup script — budget $5–$10/month.
• Monitoring. At minimum uptime checks (free via Better Stack or UptimeRobot). Production deployments want log aggregation (Axiom, Datadog, or self-hosted Grafana Loki), alerting on failed executions, and disk-usage alarms.

Add it up: a realistic production self-hosted setup is $40–$100/month at the low end and $300+/month at scale. Not free, but still usually cheaper than Cloud at equivalent volumes.

Engineering time

This is the cost nobody publishes but everyone pays. Self-hosting n8n means someone maintains it. Version upgrades (n8n ships new versions regularly — sometimes with breaking changes), security patches for the underlying Node.js runtime, database migrations, credential rotation, debugging why the queue worker stopped processing at 3am. Call it 2–4 hours per month on a healthy instance, and 20+ hours the month you hit a migration issue. At a blended engineering cost of $80/hour, that's $160–$320/month in engineering time — and spikes much higher when something breaks.

The n8n team knows this, which is why Cloud exists. The question isn't whether self-hosting is cheaper than Cloud on paper; it's whether your team is better off spending that engineering time here versus elsewhere.

Execution surprises on Cloud

If you're on Cloud Starter (2,500 executions) and one of your workflows goes into a loop because of a misconfigured trigger, you can burn through the full month's quota in an afternoon. n8n will either pause your workflows or auto-upgrade you to the next tier, depending on your account settings. Neither is great. The lesson: set execution alerts early, especially while workflows are still being dialed in.

The active-workflow ceiling

Starter caps at 5 active workflows. Pro at 15. If you're building a lot of small, focused workflows — say, one per customer or one per automation trigger — you'll hit the active-workflow ceiling long before you hit the execution ceiling. Many teams end up upgrading to Business not for executions, but because they need more active workflows. Worth knowing before you commit to an architecture.

Who n8n Is Right For

With all of that on the table, here's the honest profile of the team that n8n serves well.

Technical teams with JavaScript fluency. n8n's killer feature, beyond self-hosting, is the Code and Function nodes. You can drop raw JavaScript (or Python, via the recently-GA'd Python support) into any workflow. For a team that already thinks in code, this flexibility is addictive — you're never blocked by a missing feature, because you can always write the feature yourself.

Dev-ops-capable orgs. If you already run Docker containers, manage a few internal services, and have someone on-call for your infrastructure, adding n8n is a small marginal lift. You get a powerful automation platform for roughly the cost of its server. The economics are great.

Cost-conscious teams running high volume. If you're running more than 15,000–20,000 executions per month, n8n (either Cloud Pro / Business or self-hosted) will beat Zapier's per-task pricing by 5–20x. Our n8n vs Zapier comparison walks through the math on a realistic 5-step workflow.

Teams with data-residency or compliance requirements. Self-hosting n8n inside your own VPC means your workflow data never leaves your perimeter. For regulated industries, that's sometimes the only acceptable architecture.

Builders who want to extend the platform. n8n is MIT-licensed-adjacent (fair-code), and writing custom nodes is straightforward. If your product integrates with a niche API that no iPaaS supports, you can ship a node in an afternoon.

Who n8n Is Wrong For

This is the section most n8n write-ups skip, and it's the one that matters most if you're about to commit.

Solo founders and indie makers. n8n's ceiling is high, but its floor is also high. A solo founder trying to wire Stripe → Slack → Google Sheets is better served by Zapier or a purpose-built no-code tool. You'll ship in twenty minutes instead of two hours, and the cost delta doesn't justify the complexity.

Non-technical ops teams. If your automation builder is the ops lead who knows spreadsheets and SaaS admin but doesn't write code, n8n's node graph is genuinely intimidating. The docs are written for developers. The error messages assume API familiarity. Teams that hand n8n to a non-technical owner typically end up with one person (the only technical one) becoming the de facto bottleneck for every new workflow.

SMBs without a DevOps resource. If you picked n8n for the "free self-hosted" pitch and nobody on the team can confidently SSH into a server and debug a Docker container, you're about to learn infrastructure the hard way. Cloud is a fine out; just go in with eyes open that €24–€667/month is your actual starting price.

Teams that need AI-first automation. n8n has AI Agent nodes and LangChain integration, but it's fundamentally still an IF/THEN platform. If what you actually want is an agent that reads context and decides the next action, n8n will feel like you're fighting the paradigm. More on that below.

Workflow volumes under 500 executions per month. At low volume, Zapier's Starter tier ($29.99) is nearly identical to n8n Cloud Starter (€24). The integration breadth (7,000+ apps vs ~1,000) usually tilts the decision toward Zapier for small-volume users. n8n's math only starts winning once volume scales up.

What If You Want No-Code Without n8n's Complexity?

Here's the third-option conversation, because it's a real one in 2026.

n8n was designed before LLMs were practical. It's a node-graph IF/THEN platform — you, the builder, author the logic; the platform executes it literally. That works beautifully for deterministic workflows like "when a row is added to Sheets, enrich with Clearbit and notify Slack." It works worse for workflows that involve judgment — "triage this support email, pull relevant CRM context, decide whether to escalate."

Arahi AI takes a different shape. Instead of authoring node graphs, you give an AI agent a goal and the integrations it can use (1,500+ of them — Slack, Salesforce, HubSpot, Notion, Gmail, Stripe, and so on). The agent reads context at runtime and chooses actions dynamically. For workflows that need reasoning rather than routing, this fits the problem better than bolting an AI node onto an IF/THEN tree.

It's also simpler operationally. Pricing starts at $49/month. No self-hosting. No JavaScript. No Docker. Our n8n alternatives page breaks down the direct comparison if you want the feature-by-feature view, and the pricing page has the full tier structure.

Arahi AI isn't a drop-in replacement for n8n in every case. If you're running deterministic integration plumbing — webhook hits, transform payload, post to Slack — n8n (or Zapier) will be simpler and cheaper. If your automation needs judgment, context, or multi-step reasoning, agent-based automation is the better shape of tool.

Frequently Asked Questions

Is n8n really free?

The self-hosted Community edition is free under n8n's fair-code Sustainable Use License — but you pay for the server, the database, backups, monitoring, and engineering time. n8n Cloud starts at €24/month for a managed version. Self-hosting is genuinely cheaper at scale, but "free" is misleading at small scale where engineering time dominates.

How does n8n count executions?

Each workflow run counts as one execution, regardless of how many nodes run inside it. Sub-workflows triggered from a parent count as additional executions. Failed runs still count against your monthly quota. This model is much better than Zapier's per-task pricing for complex workflows.

What's the difference between executions and active workflows?

Executions are individual runs of a workflow. Active workflows are how many workflows can be turned on (live) at once. Starter caps at 5 active workflows and 2,500 executions; Pro at 15 active and 10,000 executions; Business is unlimited active with 40,000 executions. Many teams hit the active-workflow ceiling before the execution ceiling.

When does self-hosting n8n stop making sense?

When DevOps time exceeds the cost of Cloud, when you need SOC 2 / SSO features bundled, or when uptime and disaster recovery become business-critical. For most teams, that inflection point hits somewhere between 5,000 and 15,000 executions per month. Below that, self-hosting is usually cheaper. Above that, the engineering overhead starts to outweigh the infrastructure savings.

Is there a no-code alternative to n8n that doesn't require self-hosting or JavaScript?

Yes. Arahi AI offers 1,500+ integrations and AI agents from $49/month — fully managed, no self-hosting, no code. It replaces n8n's IF/THEN node graph with goal-directed AI agents that reason about context and choose actions dynamically. For teams that want n8n's flexibility without its operational burden, it's worth a look.