Zapier Alternatives: Five Meters, One Workflow, No Ranking

A Zapier plan tends to stop fitting at the meter rather than at the product: the hundred free tasks run out, or the bill grows faster than the work does. What happens when the allowance runs out is a separate comparison. So this page is organised around what each alternative counts, what its free plan actually allows, and which job it suits — not a ranking, because a task, a credit, an execution and a CPU second are not the same quantity and none of the pages listed at the end of this one publishes a conversion between them.

We have built the same two-step automation — a new row in Google Sheets sends an email through Gmail — in all five of these tools, on our own accounts, and the measured figures below come from those builds, each on the date and plan given at the top of this page. Every page listed at the end of this one was read on October 31, 2026.

Quick Answer

• Make — if you are willing to count operations rather than actions, and to watch what a step does to each row. Its own example shows one run costing eleven credits when a step touches ten rows, and it charges a credit for a check that finds nothing.
• n8n — if you can run a server. One execution covers the whole workflow regardless of how many steps it has, and self-hosting is free with no execution quota. There is no permanently free cloud plan.
• Pabbly Connect — if the budget is the constraint. Triggers and internal steps are exempt from the task count, and it sells by the term, from monthly up to a one-time lifetime plan.
• Latenode — if runs are short and steps are many. It bills runtime, so an extra node carries no charge of its own, though the runtime it spends does; its free plan caps a run at three minutes and gives you one worker.
• Zapier — if the automation spends most of its life waiting. Its polls cost nothing, and its help centre says so in as many words.

If what you want is the build itself rather than the meter, our Google Sheets to Gmail walkthrough is the same workflow assembled in each of the five, screen by screen.

What Each One Counts, and What the Free Plan Allows

Tool	Billing unit	Free plan	A check that finds nothing
Zapier	Task — one per successful action at the standard rate	100 tasks a month, two-step Zaps, 15-minute polling	Free. These polls never use tasks
Make	Credit — one per operation; an action module runs once per bundle	1,000 credits a month, 15-minute minimum interval	1 credit. Even when a trigger finds no new data, it still uses 1 credit for checking
n8n	Execution — one per workflow run, any number of steps	No permanent free cloud plan; self-hosted Community edition is free	Depends on the trigger node
Pabbly Connect	Task — action steps only	100 tasks a month, two-step workflows	Counted as an operation, not charged
Latenode	CPU second — compute time, not steps	10,000 CPU seconds, 5 active workflows, 10-minute trigger interval	Not published

There is no cost column, deliberately. Four different units sit in the second column and none of the pages listed at the end of this one publishes a rate between them, so a column of numbers would invite an arithmetic none of these pages supports. What the table can do is show which allowance the charge comes out of, and that is usually the thing that decides a monthly bill.

Why the Zapier Meter Runs Out

A Zapier task is any successful action the Zap performs. Triggers do not count, filters and paths do not count, editor tests do not count, and an action that errors does not count. On our own account one run of the two-step Sheets-to-Gmail Zap moved the counter from 1 to 2 — one task, for the Gmail step.

So the free plan’s hundred tasks are a hundred successful actions at the standard rate, and a two-step Zap spends one of them per run. The ceiling is not the only thing to weigh, though. The shape of it matters more. One task per action means the bill scales with how many things the workflow does, and the headline rate is not the only rate: Zapier prints 5 tasks per successful lead routed for Lead Router and 2 tasks per successful tool call for MCP, and meters Agents on a separate activity quota entirely.

Paid plans start at $19.99 a month billed yearly and $29.99 billed monthly, for Professional’s entry allowance of 750 tasks; the task slider itself starts at 100 and runs to two million. The page also carries a banner, new on the day we read it: AI steps, code, and SDK now all follow the same task-based pricing model.

Make: A Credit Per Operation, Until a Step Meets Ten Rows

Make counts operations and converts them to credits at one for one. The free plan carries 1,000 credits a month with a 15-minute minimum interval between runs. Our clean unattended run of the same workflow logged 2 operations and 2 credits — the trigger and the Gmail module.

A thousand credits a month, against Zapier’s hundred tasks, for the same work. The qualifier is what an operation is. An action module runs once per input bundle — a trigger and a search each cost one credit per run however much they return — and Make’s own worked example spells out what that costs: a Search Row module returns 10 rows for 1 credit, the Delete Rows module after it runs ten times for 10, and the scenario uses 11 credits total. One run, eleven credits. Routers, filters, Make Functions and error handlers are exempt, and Make Code is charged at 2 credits per second of execution time.

The other qualifier is waiting. Make states it plainly — Trigger modules use 1 credit per run, regardless of whether they return data — and we watched it happen: one scheduled scenario on the 15-minute minimum drew a credit per check with nothing in the sheet — four an hour, 96 a day at that interval, against a monthly 1,000. What automation costs when nothing happens works through that measurement and the other four meters beside it.

At the pricing slider’s smallest setting of 10,000 credits a month, Make lists Core at $9 billed annually and $10.59 billed monthly. Our full write-up is the Make review, and the head-to-head is Zapier vs Make.

n8n: One Execution, However Long the Workflow

n8n counts an execution as a single run of a workflow, whatever it contains. Its pricing page puts the argument in one line: Unlike other tools that charge per step or user, n8n lets you build freely and only pay when a workflow runs from start to finish. For a workflow with eight standard action steps, that is one execution against eight tasks.

The catch is the free route. There is no permanently free n8n Cloud plan — the cloud ladder starts at a trial and then €20 a month billed annually for 2,500 executions. What is free is the self-hosted Community edition. n8n applies the execution quota to paid plans, whether cloud or self-hosted, so an unpaid Community instance has none. That is a server to run, not a signup.

n8n also splits its counting rule three ways inside its own product, depending on which trigger you put at the top. A Schedule Trigger counts one execution every time the node fires, regardless of outcome; a polling node counts only when new data is found; a webhook counts every inbound request, including one with an empty body. Same product, same month, three answers.

The detail is in our n8n review, and the direct comparison is Zapier vs n8n.

Pabbly Connect: The Steps That Do Not Count

Pabbly bills action steps and exempts triggers and internal steps outright. It also keeps a second counter that it does not bill: every check made to an external app is counted as an operation, and its own FAQ quantifies the idle case — if a check is made every 5 minutes, that is 288 operations in a day, whether or not there is any new data. However, Pabbly Connect does not charge for operations.

On our own free account a test send consumed no task while a live run consumed one, which is the same line Zapier draws. The free plan card prints Get 100 Tasks, lists 2-Step Workflows among its included features, and marks Unlimited Workflows with a red cross without saying what the limit is instead. When we created workflows on a free account, three of them were created and all three stayed Active.

Pabbly also sells by the lifetime. Its pricing control offers Monthly, Yearly, 2 Yearly, 3 Yearly and Lifetime, and it opens on 3 Yearly rather than on the monthly price, so the figure on the card is not a monthly rate unless you move the control. The one-time tiers are their own decision to read carefully: the sales page and the purchase terms do not agree about the task ceiling on the largest one. Both are in the Pabbly Connect review, and the head-to-head is Zapier vs Pabbly Connect.

Latenode: Paying for Time Instead of Steps

Latenode bills CPU seconds — how long the workflow was actually executing — so an extra node carries no per-step charge the way an extra task or operation does. What it spends in runtime still counts; the vendor’s phrasing is that unlimited workflow steps do not multiply your bill. The free plan includes 10,000 CPU seconds a month and 5 active workflows, and of the five pricing pages it is the only one that converts its own allowance into a run count: up to 10,000 quick executions if runs average about 1 second.

Our run did not average about a second. The successful production run of the two-step workflow was metered at 2 CPU seconds against a duration of one second, and a separate sub-second run was still billed the one-second minimum. The vendor’s estimate and our reading are both on this page because the gap between them is the thing worth knowing.

The free plan’s other ceilings are easy to miss behind the headline: one execution worker, a three-minute limit on a run, three days of execution history, ten connected authorizations and a ten-minute trigger interval. The review is here, and the head-to-head is Zapier vs Latenode.

What Zapier Does Not Charge For

Two things, and we are paid by three of the alternatives rather than by Zapier, so take them as findings rather than as flattery.

Waiting is free, and it quantifies how free. Zapier’s help centre states it twice over: Zap triggers never use tasks, and If a Zap checks an app every two minutes, it will complete over 20 thousand polls per month. Zapier does not charge for these events. Make charges a credit for each of those checks. Pabbly also prints that it does not bill for waiting — it counts the checks and declines to charge for them. n8n’s answer depends on the trigger node. Latenode says polling does not cost extra and never says whether an empty check is an execution at all, so its answer is not published rather than known. For an automation that spends most of its life finding nothing, that is the difference between an allowance spent only on work and one that is also paying to look.

The first build asked less of us. That is an editorial judgment, not a measurement: in the two builds we put directly side by side, the mapping step is where the time went, and Zapier’s asked less of us than Latenode’s did. We are not ranking all five on it.

Where a Free Plan and Its Own Page Disagree

Three of these pages leave a free-plan reader short of the answer, in three different ways.

Latenode marks the AI Agent node as included on Free. Its comparison table carries a check mark in the Free column, the same mark it uses for the workflow builder and webhook triggers — and the Free card’s own list reads All apps, AI agents, RAG, browser, and database. On our free account the node would not run. It failed with a message about an empty Plug and Play balance, even with a model the picker marked :free. Latenode grants one Plug and Play token at registration, a fact published in its billing documentation and not on the pricing page, and neither the card nor the table mentions that the node needs one. The same table does print a grey cross for workspace invitations and the words Trial access for AI Code Copilot, so it is a table that knows how to say no.

Make’s plan comparison marks Maia as included on Free, with no trial qualifier in the row. Maia is the builder that writes a scenario from a description. On our free account, opening it returned Your free trial has ended. To continue building with Maia, upgrade to a paid plan anytime. The qualifier exists elsewhere — 30 days from signup or 45 messages, whichever comes first — just not in the row where somebody picks a plan.

Zapier puts Copilot on the free plan with an asterisk. Its own help article confirms a daily limit exists, warns at 80 per cent of it, and resets at midnight UTC — without saying what it is. The number is on neither that article nor the pricing page. We put all four builders side by side in four AI builders and their free-plan gates.

Which One to Move To

If the workflow is long and runs rarely: n8n, if you can run a server; Latenode if you cannot. Both charge for the run rather than for the steps inside it, which is the opposite of what pushed you off Zapier.

If the workflow is short and runs constantly: stay on a task meter and pick the one with the exemptions. Pabbly exempts triggers and internal steps and does not bill its operation counter; Zapier exempts triggers, filters, paths and seven of its own built-in apps.

If a single step touches many records: be careful with Make. Its eleven-credit example is its own, and it is the shape that catches people who moved on the headline allowance alone.

If the automation mostly waits: Zapier and Pabbly both print that waiting costs nothing, and a free account on either keeps all hundred of its tasks for the work. Make’s thousand credits are a larger number in a different unit, and part of them goes on the checks. Latenode does not publish an answer for an idle check, so it is not a candidate to decide this on. That is why the answer follows the shape of the work rather than a ranking.

If you want to try an AI builder without paying: go in knowing which gate you will meet. Make publishes 45 Maia messages in its Maia FAQ, though the row where you pick a plan carries only a check mark. Pabbly publishes 5 workflows and 100 Units on its AgenticAI page.

Make’s free plan is the one to open if you want to put the thousand credits against your own workflow and see what one run of it costs.

FAQ

Which Zapier alternative has the biggest free plan?

The allowances are 100 tasks, 1,000 credits, 100 tasks and 10,000 CPU seconds, and none of the pricing pages we read converts between those units, so they cannot be ranked by size. The largest figure belongs to Latenode and is in a unit none of the others uses. What can be said is what each allowance is spent on. Make documents that part of its thousand credits goes on checks that find nothing; Zapier and Pabbly document that theirs do not; Latenode does not say.

Is n8n free?

The self-hosted Community edition is, and it carries no execution quota. n8n Cloud is not — it offers a trial, then a paid ladder starting at €20 a month billed annually.

Does a lower price per operation always mean a lower bill?

Not if a step processes many records. Make’s own example puts one scenario run at eleven credits because the action module runs once per row. The per-step price is only half of the figure; the other half is how many times the step runs.

Sources

Every vendor page below was read on October 31, 2026. The measurements are from our own accounts on the dates and plans listed at the top of this page, taken for the reviews and comparisons linked above. No new test was run for this page.

• Zapier pricing — the plan cards in both billing states, and the Free card’s four features
• How is task usage measured in Zapier? — the task definition, the exemption lists, the Lead Router and MCP rates, and the polling example
• Make pricing — the Free column’s six rows and the per-credit prices
• How features use credits (Make) — the trigger rule, the module-type split and the eleven-credit example
• n8n pricing — the cloud ladder and the Community edition
• Understand executions (n8n) — the execution definition and the count by trigger type
• Pabbly Connect — the free plan row, the free-task exemptions and the operations FAQ
• Pabbly AgenticAI — the 5 workflows and 100 Units on its own free plan
• Latenode pricing — the Free vs Pay as you go table, read in a browser because its check marks and crosses do not survive automated reading
• How Latenode works — the instant and polling trigger modes
• Latenode billing — the Plug and Play token granted at registration
• How Zap triggers work — the polling interval by plan
• What is Zapier Copilot? — the daily limit that exists without a number
• Operations (Make) — what the bubble above a module counts
• Maia by Make pricing — the 30 days or 45 messages, and the per-message credit cost
• Maia by Make — the product page whose FAQ repeats the 45-message free allowance
• Choose how to use n8n — the self-hosted Community edition