Head to head · Make pricing checked 21 August 2026 · n8n 22 August · Make scheduling and billing terms 27 August
n8n vs Make: per-execution against per-step, and what self-hosting really costs
Short answer: Make charges for every step, n8n charges for the whole workflow. A fifteen-step workflow costs fifteen module runs on Make and one execution on n8n. So if your automations are long, take n8n — and if they're short, numerous, and you don't want to run a server, take Make. Neither model is cheaper in general; they're cheaper at opposite shapes of work.
Make pays us a commission if you sign up through the link on our Make review. n8n pays us nothing — and its self-hosted edition is free, so there is nothing there for anyone to pay us for. Where this page mentions Pabbly Connect further down, note that Pabbly pays us too. Read the recommendations with all of that in mind. How this works
Both are visual automation builders that connect your apps. The difference that matters is what they put on the meter. On Make, a five-step scenario costs five times a one-step scenario. On n8n, a hundred-step workflow costs exactly the same as a five-step one. If your automations are long, that isn't a discount — it's a different order of magnitude.
The short answer
| Your situation | Pick | Why |
|---|---|---|
| Long, complex, multi-step workflows | n8n | Complexity costs nothing extra |
| You don't want to run a server | Make | Hosted, with no infrastructure to think about. From $16 monthly, $12 on annual |
| Comfortable with Docker and a VPS | n8n | Self-hosted is genuinely free and unlimited |
| Many short scenarios, low volume | Make | The per-step penalty barely applies |
| Building AI agent workflows | n8n | A common default for this, with the patterns and community to match |
| You need a niche integration | Make | Make publishes the larger app list; n8n lists around 500 — check both |
The billing difference, in practice
Say you build a workflow with fifteen steps — fetch an order, look up the customer, check a condition, format some data, write to a sheet, send an email, tag the contact, and so on. It runs 500 times a month.
On Make, that's roughly 7,500 module runs, plus whatever the trigger spends polling. Make's Core tier includes 10,000 credits, so one moderately complex workflow can consume most of a plan.
On n8n, that's 500 executions. Cloud Starter includes 2,500 a month, so the same workflow uses a fifth of the allowance no matter how many steps you add to it.
The subtler benefit
Because n8n doesn't charge per step, you can build properly — add error handling, logging, retry branches, validation — without watching a meter.
On credit-based tools, good engineering practice literally costs money, so people leave it out. That's a real quality difference and it doesn't show up on any pricing page.
What each costs
| Option | Cost | What you get |
|---|---|---|
| Make free | $0 | 1,000 credits a month, 2 active scenarios. Scheduled runs no more often than every 15 minutes; webhooks fire instantly |
| Make Core | $12 | 10,000 credits, unlimited scenarios, scheduled runs down to every minute. Annual tiers $12 / $21 / $38; monthly $16 / $28 / $51 |
| n8n self-hosted | Free | All 500+ integrations, unlimited executions. You provide the server |
| n8n Cloud Starter | ~€20 annual | ~€24 monthly. 2,500 executions a month, unlimited workflows and users, 5 concurrent executions |
| n8n Cloud Pro | ~€50 annual | ~€60 monthly. 10,000 executions, unlimited workflows and users, 20 concurrent executions |
Make in US dollars. Make raised its paid tiers during August 2026 — Core moved from $9 to $12 — so any comparison still quoting $9 is out of date, and plenty are. Make's $12 is the annual rate, billed for the full year with no prorated refunds; the price you can cancel monthly is $16. n8n prices in euros: it's a German company, and the only tool we cover that doesn't bill in USD, so factor in conversion if you're budgeting in dollars or Australian dollars. n8n annual saves roughly 17%; cloud has a 14-day trial with no card, but no free cloud tier — the free option is self-hosting. We haven't confirmed n8n's cloud cancellation terms and would rather say so than assume they match Make's. n8n also removed its active-workflow caps, so older guides quoting a five-workflow Starter limit are out of date. Make checked 21 August 2026, its scheduling and billing terms 27 August; n8n 22 August.
The limit that binds is different on each — and it flipped on n8n
On Make, count the automations you'd keep switched on. The free tier allows 2 active scenarios, and for most people that ceiling arrives long before the credit count does.
On n8n, count the volume instead. Cloud plans now carry unlimited workflows and unlimited users, so executions are the only thing that scales your bill — and polling is what spends them. A workflow checking every five minutes runs about 8,640 times a month, more than three times the entire Starter allowance, from one automation that may find nothing to do.
Make has the same trap and it is worse, because Make meters the check itself. That identical five-minute schedule costs about 8,640 of Core's 10,000 credits before a single useful step runs. On the free tier, where scheduled scenarios can't run more often than every fifteen minutes, you still get roughly 2,880 checks against an allowance of 1,000 — one always-on scenario drains it before the month is half done. We're spelling that out because we earn a commission on Make and not on n8n, and it would be easy to leave only one side of this costed.
On both tools, webhook triggers fire on the event instead of polling, so they sidestep this entirely where the source app supports them. And there's a second n8n ceiling worth knowing: concurrent executions, five on Starter and twenty on Pro. Simultaneous bursts queue rather than failing, but a spiky workload can hit that well before the monthly allowance is spent. AI steps also draw on a separate credit allowance rather than the execution count, so AI-heavy workflows have two meters.
“Free” self-hosting isn't free
n8n's community edition costs nothing in licence fees and includes every integration with unlimited executions. What it costs is your attention: a server to provision, updates to apply, backups to configure, and uptime that is now your problem.
If you're comfortable with Docker and a VPS, that's an evening and then occasional maintenance — an excellent trade for unlimited automation. If you're not, you are one broken update away from silently stopped workflows and no support desk to call.
The honest question isn't whether you can self-host. It's whether you want to be responsible for it at 11pm on a Sunday.
Where Make is the better tool
- Nothing to run. It's hosted, it works, and $16 a month cancellable — $12 if you commit to a year — is still cheap for what it does
- Make publishes a larger app list than n8n's roughly 500 — check your specific apps on both vendors' directories
- Hosted is the only way Make comes, so there is no infrastructure decision to make at all — with n8n, the free path and the paid path are different things to operate
- People who have used both commonly describe Make's canvas as the more polished of the two interfaces. We haven't built in either, so that's reported rather than rated — see how we choose
Where n8n is the better tool
- Per-execution billing, which is decisive if your workflows are long
- A genuinely free self-hosted edition with no execution ceiling
- You can drop into JavaScript inline when the visual nodes run out, on any step rather than as a separate bolt-on
- Data stays on your infrastructure if that matters to you or your clients
- Cloud plans carry unlimited workflows and users, so you're never forced up a tier just to keep more automations switched on
- It is now a common default for AI agent workflows, so the patterns and community are there
Questions
Which is cheaper?
If your workflows are long, n8n, often dramatically — a fifteen-step workflow costs one execution instead of fifteen module runs. If they're short and numerous, the gap narrows and Make's entry tier is hard to beat, at $16 monthly or $12 if you commit to a year. And if your trigger polls frequently, check that cost on both before deciding: it can dwarf the plan price on either tool.
Did Make's price change recently?
Yes. Make raised its paid tiers during August 2026 — Core went from $9 to $12 a month on annual billing, with Pro and Teams rising in step. The plans themselves are unchanged. It narrows the gap slightly in n8n's favour, but it doesn't move the decision: the reason to choose n8n was always per-execution billing on long workflows, not a few dollars on the subscription. Make pays us and n8n doesn't, so check make.com yourself rather than taking our figure.
Is Make's $12 the real price?
It's the annual rate, and it commits you for twelve months with no prorated refunds if you cancel part way. Month to month, Core is $16. If you're testing whether automation sticks, pay the $16 and treat the $4 as the cost of being able to change your mind.
Did anything change on n8n's side?
The prices held, but two things underneath them moved. The active-workflow caps are gone — every cloud plan now includes unlimited workflows and users, so execution volume is the only lever, and any guide quoting a five-workflow Starter limit is out of date. And AI steps now draw on a separate monthly credit allowance rather than the execution count, which means AI-heavy workflows have two meters to watch. n8n publishes the current credit figures on its own pricing page; third-party guides get them badly wrong, so read them at the source. Re-checked 22 August 2026.
Is self-hosted n8n really free?
The licence is, and it includes all 500+ integrations with unlimited executions. You pay in server costs — a small VPS is a few dollars a month — and in your own time for setup, updates and backups. Only enterprise extras like SSO and log streaming are gated to paid tiers. Note that n8n's cloud has no free tier at all, only a trial: free means self-hosting.
Make pays you. Does that change this comparison?
It shouldn't, and the recommendation above is the test: for long workflows this page says use n8n, and the self-hosted edition we point you to costs nothing and earns us nothing. Look at what else the page tells you — that Make's headline $12 commits you for a year, and that a frequently-polling trigger can eat most of a Make plan before doing any work. Make wins here on convenience and integration coverage, which are real reasons, not commercial ones. Worth knowing that this matters more than it used to — Make and Pabbly Connect both pay us now, so n8n and Zapier, which don't, are doing more work.
Do I need to be a developer to use n8n?
Not strictly — the visual builder handles a lot without code. Two things do set it apart from Make: the free path means you run the server yourself, and the builder lets you drop into JavaScript when the nodes run out. If those read as burdens rather than features, Make will serve you better. We don't rate how hard either one is to learn, because nobody here has built in them — how we choose explains why.
Why is n8n priced in euros?
It's a German company. If you're budgeting in dollars, factor in conversion and your bank's rate — it makes direct comparisons with the US tools slightly awkward, and it's the only tool we cover that does this.
What about a one-time payment instead?
That's Pabbly Connect, which sells a lifetime licence and counts only action steps. Different trade again: cheaper forever, weaker on complex logic and error handling, and its one-time plans can't be upgraded later, so the allowance you buy is the one you keep. Note that against a cheap subscription like Make's, the payback runs to years rather than months. Disclosure, since this is a recommendation and not just a mention — Pabbly pays us a commission, as does Make, and n8n doesn't. If your workflows are long, n8n still beats both on cost and we earn nothing when you go that way.
How to decide today
Count the steps in your most important workflow. If it's under five, Make is the simpler answer and the billing model barely matters. If it's fifteen or more and runs often, n8n's per-execution model will save you real money and let you build it properly.
Then count how often the trigger checks, because on both tools that number can cost more than the work itself — and use a webhook instead of a schedule wherever the source app allows one.
Then answer the infrastructure question honestly. Try n8n's 14-day cloud trial first even if you intend to self-host — it lets you learn the builder without also debugging a server, and you can move to your own instance once you know you like it.
n8n pays us nothing. The Make review contains an affiliate link, so Make may pay us a commission if you sign up through it, and the same is true of Pabbly Connect where we mention it above. Your price is unchanged either way.