n8n Pricing 2025: What It Actually Costs to Run | 2V Automation

n8n’s pricing is one of the most-asked questions we get from clients evaluating it against Zapier or Make. The short version: there are two tracks — paid Cloud, or free self-hosted — and the right answer depends almost entirely on your execution volume and your team’s appetite for running infrastructure.

This post breaks down both tracks, the total cost of ownership including the hidden line items, and how to estimate what your specific situation will actually cost.

A note on numbers: n8n adjusts Cloud pricing periodically. We give ranges and the structure rather than fixed dollar amounts that go stale. Always check n8n.io/pricing for current tier numbers before committing.

The two pricing tracks

n8n has two ways to use it:

• n8n Cloud — managed SaaS. You sign up, log in, and start building. You pay a monthly fee that includes hosting, scaling, backups, and updates.
• n8n Self-Hosted (Community Edition) — you run the open-source Docker image on your own infrastructure. The software is free. You pay only for the VM or cluster.

Both run the same workflow engine. The difference is who operates the infrastructure.

How n8n Cloud is priced

n8n Cloud is structured into multiple tiers (Starter, Pro, and Enterprise, with the names and tier counts having shifted over time). Each tier bundles:

• A monthly cap on workflow executions
• A maximum number of active workflows
• Feature access (multi-user, environments, SSO, audit logs, log streaming, external secrets, etc., gated by tier)

The pricing model — and the part that’s structurally important — is per workflow execution, not per task or per operation. A workflow that processes 1,000 records in a single run counts as one execution.

This is the single biggest economic difference between n8n and Zapier. On Zapier, that same workflow would be 1,000 tasks. At any meaningful volume, the per-execution model is dramatically cheaper.

Ballpark for 2026:

• Starter tier: low double-digit dollars per month. Caps active workflows in the single digits and executions in the low thousands. Good for teams just testing the platform.
• Pro tier: low-to-mid hundreds per month. Higher execution caps, more active workflows, multi-user.
• Enterprise tier: custom pricing. SSO, audit logs, log streaming, external secrets management, multiple environments, dedicated support, and usually significantly higher execution caps. Used by teams running n8n as production infrastructure.

Check the live page for current numbers — the tier structure and price points do update.

How self-hosted pricing works

Self-hosted n8n has no software fee. The Sustainable Use License (which replaced n8n’s earlier license in late 2022) lets you use, modify, and host the code for any internal business purpose at no cost. The only thing the license restricts is reselling n8n as a hosted product to others.

So your “n8n cost” self-hosted is really just the cost of the infrastructure you run it on. That breaks down to:

• Compute (your VM or cluster). A single VM with 2 vCPU and 4 GB of RAM runs n8n comfortably for most teams. Common pricing on AWS, DigitalOcean, Hetzner, or any reasonable cloud: $10–$30/month.
• Database. Postgres for production. Either a small managed Postgres instance ($15–$50/month on RDS, DigitalOcean Managed DB, etc.) or self-managed on the same VM (free, but you own the backups).
• Redis (for queue mode at scale). $10–$20/month for a small managed Redis instance, or free if you self-manage.
• Storage and backups. Cheap — a few dollars a month even at modest volume.
• Bandwidth. Negligible for most workloads.

Total for a small-to-medium production setup: $30–$100/month, all in.

For high-volume setups with queue mode, multiple workers, and proper managed services: $200–$500/month is typical. We’ve seen client setups handling millions of executions a month running comfortably under $300/month in infrastructure cost.

Hidden costs: what most pricing pages skip

Software fees aren’t the whole story. The real total cost of ownership includes:

On Cloud

• Premium app access tiers. If your workflows touch apps that are gated to higher Cloud tiers, you pay for those tiers regardless of execution volume.
• Overage charges. Going past your execution cap triggers per-execution overage billing. Estimate volume conservatively or stay one tier above your peak.
• Workflow seat growth. Adding team members usually means going up a tier.

On Self-Hosted

• Time to maintain. Someone updates the Docker image, monitors the Redis queue depth, watches Postgres performance, applies security patches. For a team that already runs services, this is a couple of hours a month at most. For a team without ops capacity, it’s a real cost.
• Backups and disaster recovery. Postgres dumps, encryption key backups, snapshot policies. Trivial in time, important in dollars-saved-when-things-go-wrong.
• Migration cost if you switch hosting. If you start self-hosted and want to move to Cloud (or vice versa), it takes a day of work to migrate workflows and credentials. Plan for it.
• The encryption key. Lose it and your credentials are gone. We’ve seen this happen to teams who didn’t realize it lived only on the original instance. Back it up properly.

Both tracks

• Model API costs (for AI workflows). If you’re using n8n’s AI nodes — OpenAI, Anthropic, Gemini — you pay those providers directly. Typically $50–$500/month for a moderate-volume business automation; can run higher for high-volume customer-facing workflows.
• Maintenance and improvement effort. Workflows need attention as your business changes. APIs deprecate. Business logic shifts. Plan for ongoing engineering time or a retainer relationship.

n8n Cloud vs Zapier on cost

The single biggest reason to look at n8n is cost at volume. Concrete numbers:

A few caveats. The “tasks vs executions” mapping is rough — it depends entirely on how your workflows are structured. A Zapier task is a single action; an n8n execution is a full workflow run that may touch dozens or hundreds of records. We’ve assumed an average of 100 records per execution, which is conservative.

The pattern is clear: at low volume, Zapier and n8n Cloud are roughly comparable. As volume grows, Zapier’s per-task pricing pulls ahead in cost while n8n’s per-execution pricing stays flat. By the time you’re at hundreds of thousands of tasks a month, n8n Cloud is half to a quarter of the Zapier bill — and self-hosted is a fraction of either.

Run your specific numbers on the workflow cost calculator.

When Cloud is the right choice

Pick Cloud if:

• You don’t have ops capacity to run a Node.js service in production
• Your execution volume is low enough that Cloud is cheap (under a few thousand executions/month)
• You don’t have data residency or compliance constraints that require on-prem
• You want zero overhead — sign up, build, never think about infrastructure
• You’re at the validation stage and want to keep moving fast

The trade is a higher per-execution price for not having to operate anything.

When self-hosted is the right choice

Pick self-hosted if:

• Your execution volume is high enough that Cloud pricing gets uncomfortable
• You have data that can’t leave your infrastructure (regulated industries, customer PII at scale, internal systems behind a firewall)
• You want to write custom nodes against internal libraries
• You already operate other services — one more is marginal cost
• You’re building serious AI workflows and want to integrate with internal models or vector stores you host yourself

The trade is operational responsibility in exchange for lower cost and more control.

A simple decision framework

1. Are you under 1,000 executions a month? Use Cloud Starter. The cost isn’t there yet to justify self-hosting effort.
2. Are you 1,000–10,000 executions a month? Either works. Self-hosted saves money but only if you’ll actually operate it. Default to Cloud unless you have an ops team.
3. Are you over 10,000 executions a month? Self-hosted starts being meaningfully cheaper. If you have ops capacity, go self-hosted. If not, Cloud’s higher tiers still make sense compared to Zapier-style alternatives.
4. Do you have data residency, compliance, or air-gapped requirements? Self-hosted, regardless of volume.
5. Are you building serious AI workflows at scale? Self-hosted gives you more flexibility on model hosting and vector DB integration. Either works otherwise.

How to estimate your bill

Three numbers tell you most of what you need to know:

1. Workflow executions per month. Count how many times your workflows will run end-to-end. A workflow that runs every hour is ~720 executions/month. A workflow that runs once per inbound webhook scales with your traffic.
2. Active workflow count. How many distinct workflows you’re going to have live simultaneously.
3. AI model usage (if applicable). Estimated tokens per execution × executions per month × your model’s price. Roughly: GPT-4-class models cost cents per call; smaller and open-weights models cost fractions of a cent.

Plug those into the Cloud tier pages or estimate self-hosted infrastructure from your provider’s calculator. Add 20–30% margin for growth and unexpected volume. That’s your number.

Or, if you’d rather not do the math: our workflow cost calculator lets you input your specific numbers and compares Cloud vs self-hosted vs Zapier vs Make side-by-side.

Total cost of ownership: a realistic example

A client we worked with last year was paying $1,800/month on Zapier for what amounted to twelve workflows handling ~400,000 tasks/month. Their actual workflow count was small; the task count was high because most workflows looped over arrays of records.

We migrated them to self-hosted n8n. The numbers:

• One-time migration: ~2 weeks of engineering work (this scales with your specific complexity; theirs was moderate)
• Infrastructure: $120/month (managed VM, managed Postgres, managed Redis)
• Model API costs (added two AI-driven workflows during migration): ~$200/month
• Maintenance retainer: $1,000/month

Their new fully-loaded total: $1,320/month — meaningfully less than the old Zapier bill, with significantly more capability (real branching, error workflows, AI agents, code-based logic). Payback on the migration was under 6 months.

This is a typical shape. The savings come not from the software being cheaper alone — they come from removing the per-task pricing tax at scale.

Related reading

• n8n automation: the complete guide — the long-form pillar covering everything else
• What is n8n? — the basics
• n8n vs Zapier: the complete comparison — the head-to-head
• n8n vs Make vs Zapier: which automation tool wins? — the three-way
• How to migrate from Zapier or Make to n8n
• Workflow cost calculator
• Automation ROI calculator

If you’re trying to figure out the right path — Cloud, self-hosted, or migrating away from another tool entirely — and where automation will actually pay back in your business, the fastest answer is our Efficiency Scorecard. 15 minutes. Free. You keep the output regardless.