Make.com vs n8n: billing, hosting, and limits compared

A workflow with six steps runs 200 times in a month. On n8n that is 200 executions. On Make it is about 1,200 credits, because Make counts each module run. Both numbers are small. Multiply the volume by 50 and the gap is the whole pricing story, which is why a table of headline plan prices tells you very little.

This post compares Make.com and n8n on the things that decide the choice: how each one counts usage, where it runs, what the plans cap, and who has to maintain it. Prices were read from the vendors' own pages on October 5, 2026. Plans change, so treat the figures as a dated snapshot.

One disclosure first. Gonzo Digital builds Python agents, not Make or n8n flows, so we earn nothing from either choice. Where a simple flow is the right answer, we say so.

The short answer

Choose Make when nobody on your team wants to run a server, your flows connect common apps, and you would rather drag modules on a canvas than read code. Choose n8n when you want to self-host, need code steps in the middle of a flow, or run long workflows where counting every step would get expensive.

Neither is better in general. The rest of this post gives you the numbers to check against your own flows.

What Make.com is

Make is a hosted visual builder. You draw "scenarios" as a flow of modules, and Make runs them on its own servers. Make says it offers 3,000+ pre-built apps.^[1] It is the product formerly called Integromat; Make's own announcement calls Integromat a legacy product that it stopped supporting in 2023.^[2]

Its billing unit is the credit. For non-AI apps, one operation equals one credit.^[3] An operation is one run of one module, and a module that processes ten items runs ten times.^[4] A trigger that checks for new data counts once per check, however many items it returns.^[4] Routers and filters do not consume credits.^[5]

Make also lists AI agents and an MCP server in its help center, which lets assistants such as Claude and ChatGPT run scenarios in your account.^[6]

What n8n is

n8n is a workflow tool you can use in two ways: pay n8n to host it, or run it yourself. n8n says it has over 500 integrations, plus HTTP and custom API requests for everything else.^[7] Most of its source code sits on GitHub under the Sustainable Use License, a license that n8n says is based on the fair-code model.^[8]

Its billing unit is the execution: "a single run of your entire workflow," however many steps it has and however much data it processes.^[9] Every cloud plan lists JavaScript and Python code steps, custom API requests, and an AI Agent node.^[9]

Side by side

Sources: Make's pricing page and help center, n8n's pricing page and home page.^{[1][3][7][9][10]} Make's pricing page refused a direct request from our script, so its figures were read through a page-reading tool. Check them on the page before you decide. Make also showed annual billing as "save 15% or more" without a fixed figure we could quote.^[10]

What the units do to your bill

Take a lead-intake flow with a webhook trigger and five more steps, run 1,000 times a month. On n8n that is 1,000 executions, which fits inside the 2,500 of the Starter plan.^[9] On Make, each run is about six module runs, so about 6,000 credits, assuming every module runs once per lead.^[3][4] That is above the 5,000-credit entry tier, so you would move up a tier; read that tier's price on Make's page.^[10]

Now flip it. A three-step flow that runs 100 times a month is about 300 credits on Make, which fits in the free plan's 1,000.^[10] On n8n cloud the same flow costs at least EUR 20 a month, unless you host it yourself.^[9] At low volume with few steps, Make is cheaper. As flows get longer or handle batches, n8n's unit gets better.

Loops matter most. Make runs a module once per item, so a step that handles 500 rows is 500 operations.^[4] n8n's page says it does not matter how much data one execution processes.^[9] If your flows loop over large lists, run the arithmetic on your real flow, not an average one.

Two cautions. First, n8n's cloud plans cap more than executions: Starter allows 5 concurrent executions and a maximum execution time of 5 minutes, and Pro allows 40 minutes.^[9] Second, n8n's Business plan, the EUR 667 a month tier for 40,000 executions, is listed as self-hosted only as of October 5, 2026. The page says the hosted version has a wait list.^[9]

Self-hosting is free in license fees, not in time

n8n's Community edition costs nothing to download, and a self-hosted install without a license key runs as that free edition.^[11] You still provide the server, the updates, the backups, and the security. n8n's docs list several install routes, including Docker and cloud providers.^[11] For a team with an engineer who enjoys this, that is fine. For a team without one, count the hours before you count the savings.

Self-hosting is also the way to keep data on infrastructure you pick. n8n's hosted plans store data in the EU, on servers in Germany.^[9] If a rule or a contract says where customer data may sit, that is a real reason to choose n8n, but it does not settle the legal question. Read the rule, then the vendor's data terms.

The license deserves a second look if you are an agency or an independent consultant. The Sustainable Use License lets you use and modify n8n for your own internal business purposes, or for non-commercial use, and restricts passing it to others except free of charge for non-commercial purposes.^[8] Running an n8n instance for a client's business is a case to check against n8n's license FAQ, or with license@n8n.io, before you build.^[8]

When Make is the better fit

• Nobody on the team wants to run infrastructure, and a hosted service is the point.
• Your flows connect common SaaS apps that Make already lists.
• Flows are short and volume is modest, so the credit count stays low.
• You want to see the whole flow as a canvas and hand it to a non-technical colleague.

When n8n is the better fit

• You need to self-host, or you want data to stay on infrastructure you control.
• Flows are long, branch a lot, or process big batches, so counting each step would add up.
• You want code steps in the middle of a flow without leaving the tool.
• Someone on the team is comfortable maintaining a server, or you are happy to pay n8n to host.

Using both

It is common to run both: a hosted tool for light, client-facing flows, and a self-hosted one for internal data work. The cost is two tools to learn and two places to look when something breaks. Neither vendor's pages that we read describe a converter between the two, so a move from one to the other is a rebuild. Check with each vendor before you rely on that.

When a plain flow stops being enough

A Make or n8n flow is a good answer for a straight line: a form submission becomes a sheet row, a Slack message, and an email. If that is your whole need, build it yourself and skip us.

Trouble starts when the task needs judgment and memory. A first reply that reads the inquiry, answers in the customer's language, remembers the earlier conversation, and hands off to a person when it should is hard to keep tidy as a canvas of modules. That is where we build Python agents instead. The Lead Response Engine and Client Onboarding Pipeline are examples, and our comparison of DIY tools against a built system lays out the trade with the same dated vendor prices. We do not build or host flows in either tool, but our pages on n8n and Make say where each one's limits start. Our Zapier vs Make comparison covers the third big tool.

Questions people ask

Is Make cheaper than n8n? At low volume and with short flows, usually yes, because Make's free plan covers 1,000 credits. At higher volume, or with long flows and loops, n8n's per-execution unit can come out ahead. Run your own flow through both units, as in the example above.

Is self-hosted n8n free to use in a business? The software costs nothing, and the license allows use for your own internal business purposes.^[8] You pay for the server and the time to run it. Offering n8n to others as a service is a different case; ask n8n.

Can I move workflows between Make and n8n? We found no official converter in the pages we read. Plan for a rebuild.

Which has more integrations? By the vendors' own figures, Make lists 3,000+ pre-built apps and n8n lists 500+ plus HTTP requests.^[1][7] Counts are not like for like, so check the specific apps you need on each site.