n8n Pricing vs Zapier and Make (2026): Cost per Workflow
n8n pricing starts at €20 per month for 2,500 Cloud executions, billed annually, or you can self-host the Community Edition and pay for infrastructure and operations. Zapier bills in tasks. Make bills in credits, which replaced “operations” as the purchased unit. Those units are why a simple plan-price comparison cannot tell you what your workflow will cost.
Last checked: 28 September 2026. Prices below are the vendors’ public starting prices in their displayed currency and billing term, not quotes for a specific workload. Recheck the linked pricing pages before buying.
What each platform counts
| Platform | Billing unit | What to count in your workflow |
|---|---|---|
| Zapier | Tasks | Successful action steps and connector calls; rates can vary for AI, code, and connector types |
| Make | Credits | Module actions; standard non-AI actions generally use one credit, while some advanced features use more |
| n8n Cloud | Executions | A complete workflow run, regardless of node count |
| n8n Community Edition | Server and operations | Compute, storage, backups, monitoring, and the engineer responsible for them |
A trigger is not the same as a billable action. For example, Zapier says polling for new data does not itself use tasks; a successful action after the poll can. Make’s credits guide explains how an operation maps to a credit, including exceptions for AI features. n8n’s pricing FAQ defines one execution as a whole workflow run.
Public plan prices in September 2026
| Platform | Public starting point | Usage shown by the vendor |
|---|---|---|
| Zapier Free | $0 | 100 tasks/month |
| Zapier Professional | From $19.99/month | Select a task tier and billing term |
| Zapier Team | From $69/month | Select a task tier and billing term |
| Zapier Enterprise | Quote | Custom terms |
| Make Free | $0 | 1,000 credits/month |
| Make Core | $12/month at the displayed monthly setting | 10,000 credits/month |
| Make Pro | $21/month at the displayed monthly setting | 10,000 credits/month |
| Make Teams | $38/month at the displayed monthly setting | 10,000 credits/month |
| n8n Cloud Starter | €20/month, billed annually | 2,500 executions/month |
| n8n Cloud Pro | €50/month, billed annually | 10,000 executions/month |
| n8n Community Edition | No software subscription | Self-hosted; infrastructure and operations extra |
Sources: Zapier pricing, Make pricing, and n8n pricing. Zapier’s price changes with the task tier and billing term; Make’s price changes with credits and billing term. The table is a starting point, not a cross-currency total-cost ranking.
n8n also sells a paid self-hosted Business tier for collaboration and governance features. Do not assume every feature on the Business pricing page is in the free Community Edition. On n8n Cloud, the current Starter and Pro pages list unlimited users and workflows; the old 5/15-workflow limits no longer belong in a 2026 comparison.
Same workflow, three different meters
Suppose 1,000 orders a month each trigger a workflow with four external actions. In Make, assume the same logic uses five standard module actions. Ignore AI-provider usage for this base case.
| Monthly runs | Zapier estimate | Make estimate | n8n Cloud estimate |
|---|---|---|---|
| 1,000 | 4,000 tasks if each action uses one task | 5,000 credits if each module uses one credit | 1,000 executions |
| 5,000 | 20,000 tasks | 25,000 credits | 5,000 executions |
| 50,000 | 200,000 tasks | 250,000 credits | 50,000 executions |
At 1,000 runs, Make Core’s displayed 10,000-credit allowance and n8n Cloud Starter’s 2,500-execution allowance cover this simple example. At 5,000 runs, the example exceeds Make Core’s 10,000 credits and n8n Starter’s 2,500 executions; n8n Pro’s displayed 10,000 executions cover it. For Zapier, choose a task tier that covers the actual measured tasks and read its live price. These are capacity checks, not claims that one product is always cheapest.
The formula to use on your own workflow is:
monthly units =
monthly trigger events
× billable actions or module actions per event
× measured fan-out and retry factor
For n8n Cloud, replace the action count with executions per event. A single order may trigger more than one execution if your design starts sub-workflows separately. For Make, AI-provider credits can exceed the one-credit-per-module baseline. For Zapier, built-in steps and variable-rate actions mean a straight step count is only an estimate.
Hidden costs to measure before switching
Loops and fan-out. A Make scenario that processes 20 line items through three standard modules can use about 60 credits in that branch. Zapier’s action count depends on how the loop is built. n8n Cloud still counts the workflow run, but larger runs can demand more memory and execution time.
Retries and error routes. Record what actually executes after a downstream API fails. Make’s router and basic error-handler control modules do not consume credits, but downstream actions can. Zapier charges for successful action work, so a failed poll is not a task. n8n execution counts can rise if your retry design starts new workflows. Measure the platform’s usage log rather than applying a blanket 20% surcharge.
AI and code. Make’s managed AI features can use credits beyond one per module; third-party AI APIs may also bill tokens separately. Zapier now has task rates for AI steps, code, and SDK calls. n8n’s workflow execution fee does not include the price of the LLM API called inside the workflow.
Self-hosting. A Community Edition server needs backups, patching, monitoring, queue capacity, TLS, and an owner for incidents. Paying only the VPS invoice is not the total cost. If webhook delivery matters, confirm the sender’s retry behavior and put a durable queue in front of an instance that may restart. The n8n self-hosting guide covers the operating work.
Which choice fits
- Start with Zapier when speed and connector coverage matter most and your measured task count fits a tier you accept. Its public Free plan handles 100 tasks per month; paid plans add multi-step workflows.
- Start with Make when a visual builder and managed operations matter, and your measured module and AI credit use fits a selected credit tier. Make now has code and AI features, so evaluate those on their own credit rules.
- Choose n8n Cloud when whole-workflow execution pricing fits your run volume and you want n8n without running the service yourself.
- Choose self-hosted n8n when you need control of the instance and have an operations owner. Check whether Community Edition has the governance features you need before treating the license as free.
The comparison is a decision about maintenance as well as invoices. A managed platform handles service uptime; a self-hosted instance gives you control and the responsibility that comes with it. For production workloads, I look at retries, approvals, secrets, and recovery before optimizing the last few cents per run. The Zapier-to-n8n migration guide covers the implementation choices.
Pricing check you can run in 15 minutes
- Pick one representative workflow and count its successful external actions, Make module actions, and n8n executions over a real week.
- Include line-item fan-out, AI or code usage, and any replayed work after failures.
- Multiply by four for a monthly starting estimate, then check the busy week’s volume as a peak case.
- Select the capacity tier on each Zapier, Make, and n8n pricing page. Record currency and billing term.
- Add implementation time and, for self-hosting, a named owner for backup and incident response.
That gives you a budget you can defend. If the workflow is core product logic, has strict latency needs, or needs transaction-level guarantees, price a small code service alongside the visual platforms. A low subscription price does not repair an unreliable architecture.
Changelog
- 2026-09-28: updated Make’s billing unit to credits, replaced obsolete Zapier plan names and prices, refreshed n8n plans, and rebuilt the comparison around measurable workflow units.
- 2026-07-11: merged the Make vs n8n production-workloads guide into this page.
Running, migrating, or hardening n8n in production: n8n consulting.
Hour estimates per phase, or a clear no with reasons.
Your agents answer from whatever the retriever finds, and too often that is last quarter's truth. I build the context layer they answer and act from: a temporal knowledge graph that keeps every fact with its source and the time it held, reads with each person's own permissions, and writes nothing without a person's approval. On your own tenant, billed by the hour, step by step.