n8n vs Zapier vs Make: which automation tool costs less for your workflows?

n8n, Zapier and Make all connect apps and run automated workflows, and all three now sell agents on top. The real difference is how they count what you use: n8n charges per workflow run, Zapier per action that completes, and Make per module action. The same automation can therefore cost very different amounts on each. This comparison sets out their plans, the billing unit behind each price, a worked example in each unit, their model features and where each one runs, from the three pricing pages as published on October 9, 2026. It is a documentary comparison: we did not build the same workflow on all three or measure their speed.

n8n vs Zapier vs Make in one table

Table: n8n, Zapier and Make compared (pricing pages checked October 9, 2026; Make's again on November 3)

n8n's prices appeared in euros when we checked from Spain, while Zapier and Make showed US dollars. Zapier and Make price their plans by volume: the figures above are the entry points, and both pages let you pick a higher monthly volume at a higher price.

How each one counts usage

n8n: executions. n8n says its pricing is "based on monthly workflow executions, regardless of complexity", and that you "only pay when a workflow runs from start to finish". A workflow with 2 steps and one with 40 steps cost the same per run. Every plan includes unlimited users, unlimited workflows and every integration; the plans differ in executions, concurrency (5 runs at once on Starter, up to 50 on Pro) and features such as shared projects, SSO and Git version control.

Zapier: tasks. In Zapier's words, tasks are used "when a Zap successfully moves data or completes an action for you automatically". Checking for new data (polling) never uses a task, and the built-in Filter, Formatter, Paths, Delay, Looping, Tables and Forms steps are free. Some apps or actions use more than one task, and Zapier now bills model steps, code and its SDK from the same task pool at their own rates. A Zapier MCP tool call uses two tasks. When you reach your limit, Zapier can switch you to pay-per-task billing at a higher rate, or pause your Zaps if you turn that off.

Make: credits. Make says each module action "like adding a Google Sheet row or fetching Gmail account data, counts as one credit", including reading data from an app or webhook, searching, creating, updating, deleting, transforming, aggregating and iterating. Most actions use 1 credit, some features of Make's own model provider use more, and running custom code costs 2 credits per second. If you run out, scenarios stop until you add credits, which you can buy in bundles of 1,000 or 10,000.

The same workflow priced three ways

Take a lead-handling automation: a form submission triggers it, then 5 steps run (summarize the lead with a model, create a CRM contact, post to Slack, add a spreadsheet row, send a confirmation email). It runs 1,000 times a month.

• n8n: 1,000 executions. That fits the Starter plan's 2,500, at €24 a month billed monthly.
• Zapier: 5 actions × 1,000 runs = 5,000 tasks, if every action uses one task (the model step may use more). That is 50 times the free plan's 100 tasks, so you need a Professional tier sized for at least 5,000 tasks, priced on Zapier's slider above the $29.99 entry price.
• Make: the trigger module plus 5 actions = 6 credits per run, so about 6,000 credits a month. That fits the Core plan's 10,000 credits at $10.59 a month ($9 billed annually).

Now double the workflow to 10 steps. n8n still counts 1,000 executions; Zapier counts about 10,000 tasks and Make about 11,000 credits, just over Core's 10,000. That is the core of the choice: n8n gets relatively cheaper as workflows grow longer, while Zapier and Make get relatively cheaper for short workflows that run very often. The model step itself also costs money on the model's side if you bring your own API key; our guides to OpenAI API pricing and the LLM API price comparison show what a call costs.

Agents and MCP on each platform

All three sell model features, and all three support the Model Context Protocol, the standard we explain in what MCP is:

• n8n includes agents, MCP and model nodes and models that work without your own API keys from its Starter plan. Its feature table lists an Agent node, an MCP Server Trigger and MCP Client and human approval for tool calls. Its built-in Assistant uses monthly credits (1,600 on Starter, up to 9,600 on Pro).
• Zapier lists model fields on Professional, Zapier Copilot from the free plan, separate Agents and Chatbots add-ons, the Zapier SDK to connect your own agents to "9,000+ apps", and Zapier MCP so that chatbots can act in your apps.
• Make offers Maia, which builds automations and agents from a conversation, Make Agents and web search (both in beta), an MCP server, a content extractor and a model toolkit that can use Make's own model provider or your own model key.

An agent decides its own steps, so its usage is harder to predict than a fixed workflow: on Zapier and Make every extra step it takes is another task or credit, while on n8n it stays inside one execution (model calls through n8n's own model gateway use separate credits). Before you let one loose on customer data, read what agents can and cannot do for a business.

Hosting, data and limits

• n8n is the only one of the three you can run on your own server. Its Community Edition is free on GitHub, and its Business plan (€800 a month billed monthly, 40,000 executions) is self-hosted only. Self-hosting means your data stays on your infrastructure, but you maintain it.
• Make lists AWS in the EU or North America as its hosting on every plan, which matters if you need to keep data in Europe. Its free plan limits you to 2 active scenarios, a 15-minute minimum interval between runs and 5-minute executions.
• Zapier's pricing page lists no self-hosted plan. Enterprise adds annual task limits, observability and advanced deployment options.

Execution limits also differ: a single run can last up to 5 minutes on n8n Starter and Make Free, and up to 40 minutes on n8n Pro and Make's paid plans.

Which one should you choose?

1. Choose Zapier if the people building automations are not technical, you need an app that only Zapier supports, or your workflows are short. Its free plan is enough to test a two-step Zap.
2. Choose Make if you want a visual builder at a low price per action and your workflows run often but stay short to medium length. Its free 1,000 credits are enough for a real test.
3. Choose n8n if your workflows have many steps, you need code steps or self-hosting, or you want unlimited users without paying per seat. A developer will set it up faster than a marketing team.
4. Count your own units before you pay. Write down how many runs a month you expect and how many actions each run has, then convert them into executions, tasks and credits as in the example above.

Prices and billing units change: Zapier has just moved model steps and code onto its task pool, and Make's FAQ says credits became its billing unit on August 27. This page is reviewed every quarter.