Latenode vs Pabbly Connect: One Counts Time, One Counts Actions

Put the same workflow into Latenode and into Pabbly Connect and the bill arrives for different reasons. Latenode charges for how long the run takes. Pabbly charges for how many times the run touches an app that is not Pabbly. Neither is hiding this — both state it on their own pages — but the consequence only shows up when you look at a specific workflow, because the shape of what you are building moves the two bills in different directions. Volume still matters on both — Latenode’s rate steps down as monthly CPU seconds rise, and Pabbly’s paid tiers carry different task allowances — but shape is the part that decides which meter you run into first.

Quick Answer

Here is what each one counts, in its own words, and what its free plan gives you.

	Latenode	Pabbly Connect
What it counts	Compute time — total compute time for the run, not the number of nodes or operations	External actions — if a action is being performed on an external app, it will be considered as a task
Free allowance	10,000 CPU seconds a month	100 tasks a month
Minimum per run	1 CPU second, billed in 1-second steps	Not stated; a task is per action
Steps per workflow, free	Unlimited — steps do not multiply your bill	2-Step Workflows
Run length, free	3 minutes per run	Nothing printed either way
Internal steps	Counted — they take time	Not counted, on any plan
Idle polling	Not stated on the page that defines CPU seconds	Counted as an operation; operations are unlimited
Card to start	Not required	Not required

Read the lower rows before the upper ones. The allowances are not comparable — a CPU second and a task measure different things, and neither of the pages that define them mentions the other’s unit — but the limits underneath are, and they point in opposite directions.

Latenode Charges for Time

Latenode’s documentation is direct about it: CPU seconds are consumed during a scenario execution (from start to finish of the run). Billing is based on total compute time for the run, not on the number of nodes or operations. The same page says the node count does not change the rate: A scenario with 10 nodes and one with 2 nodes that both finish in a few seconds are each charged at least 1 CPU second on Free.

That floor is the part worth internalising. On the Free plan, usage is billed in 1-second steps. Minimum billable execution time is 1 CPU second per run. Nothing is cheaper than one CPU second, however fast it finishes. Note that billed CPU seconds and the duration a run reports are not the same measurement, and our own readings show them diverging.

Our own runs matched it, and one of them sat slightly above it. On September 26, 2026, a two-step production run on a free account finished in one second and was metered at two CPU seconds. On October 25, 2026, a sub-second run was billed the one-second minimum. The second reading is the documented floor. The first is one run on one day, on a free account with an instant trigger, and we did not keep a screenshot of it — so treat it as a reading rather than a rate.

Two things are explicitly free, and the first is narrower than it sounds. Running a single node does not consume CPU seconds, so you can test and configure nodes on the canvas for free — that is one node at a time, not a full scenario run used as a test. And when a scenario calls a nested Latenode scenario, CPU seconds are counted for the main scenario’s execution time only.

Pabbly Charges for Reaching Outside

Pabbly’s unit is narrower than it first sounds. The tooltip on its plan cards says a task is counted when a action is being performed on an external app. The steps that are not external are named, and they are free on every plan: Unlike others we don’t count tasks for internal apps like Filter or Router or Text Formatter or Number Formatter or Schedule or Iterator or Date/Time formatted or Email Parser or JSON Extractor or Array Functions.

So a Pabbly workflow can be long without being expensive. Add five formatters and a router and the task count does not move. Add one more action against an outside app and it does — and two actions against the same app are two tasks, because the unit is the action rather than the app.

Waiting is free too, and stated as such: Every check made to an external application to see if there is any new data in the entire day is counted towards the operation — and operations are listed as unlimited on every card. A workflow that polls all day and finds nothing does not spend from the hundred; what it spends is whatever it does once it finds something.

One test we ran cost nothing. On October 25, 2026, a test send on a free account consumed no task, and the live run that followed consumed one. That is one workflow on one day, not a rule the page states.

What Latenode Does Not Say About Waiting

The page that defines CPU seconds covers what a run costs. It does not say what an idle polling check costs, and we are not going to fill that in from the shape of the rest of the page.

This matters because it is the one axis where Pabbly’s answer is unambiguous and Latenode’s is absent. If your workflow spends most of its life waiting for something to happen, you can price that on Pabbly from its own page and you cannot price it on Latenode from the page that otherwise tells you everything about CPU seconds.

What idling costs across five tools, including these two, is the subject of our separate look at what automation costs when nothing happens.

The Free Plans Restrict Opposite Things

Latenode says the quiet part on its own pricing page: Unlimited workflow steps do not multiply your bill. What it caps instead is time and breadth: three minutes per run, five active workflows, one parallel worker, three days of run history. The paid tier lifts each of those — the page states them as Unlimited active scenarios (vs 5 on Free), 10 min per run (vs 3 min), 30-day run history (vs 3 days).

Pabbly’s free card does the reverse. It says 2-Step Workflows, and nothing on the pages we read states a run-length limit or says there is none. A trigger and one action is the whole budget.

Put concretely: a five-step workflow that finishes instantly is free on Latenode and needs a paid plan on Pabbly. A two-step workflow that runs for four minutes will not finish inside Latenode’s free ceiling; whether Pabbly has a ceiling of its own is not something its pages told us.

The three-step case — a form, a spreadsheet and an email — is where Pabbly’s two-step ceiling bites first.

Two Things That Did Not Line Up

Both tools pay us, so this section exists for both of them rather than for neither.

On Latenode, on October 25, 2026, the CPU-second figure in the sidebar and the figure on the statistics page were not the same number on the same free account. The two are separate displays and we did not establish that they were reporting the same quantity, so this is two screens disagreeing rather than a product contradicting its documentation. Which one the allowance is drawn against is not something we could settle.

On Pabbly, on October 29, 2026, we created three workflows on a free account and all three stayed Active, against a pricing card that offers up to two. Nothing stopped the third at creation time. We are not treating that as a documented allowance — the card says what it says — but if you are planning around the number two, plan around the card and not around what the product let us do.

One Latenode Figure That Does Not Travel

If someone quotes you a Latenode number from their own account, check which unit it is in. The documentation notes that Some older accounts still see credits instead of CPU seconds in the UI. On those accounts, 1 credit equals 30 CPU seconds.

A figure in credits and a figure in CPU seconds differ by a factor of thirty, and neither screen labels itself as the legacy one in the quote you were sent. This is the sort of thing that makes two people comparing notes disagree about a product that is behaving consistently for both of them.

What the Paid Side Looks Like

Latenode has no base fee. Past the free 10,000 CPU seconds you pay for runtime on a descending scale, starting at $0.00012 per CPU second for the next ninety thousand and falling as the volume rises. Runtime is not the only line, though: Plug and Play tokens are a dollar each and are billed on top of the CPU seconds a run uses.

Pabbly sells both recurring plans and one-time lifetime ones. The recurring tiers carry task allowances and their prices move with the billing period you select on the page; the lifetime tiers are a single payment with their own monthly task figures and their own workflow limits. We cover the way that selector changes what you see, and the terms that govern task expiry, in our Zapier versus Pabbly comparison.

Verdict: the Shape Decides, Not the Size

Count the actions your workflow performs against outside apps, then estimate how long one run takes. Those two numbers tell you which meter you will reach first; how many times a month you run it decides what that costs.

Pabbly fits a workflow that performs few actions against outside apps but does a lot of work between them, or one that mostly waits. Filters, routers and formatters are free, polling is free, and a long quiet workflow draws almost nothing from the hundred tasks. The free plan’s two-step ceiling is the thing that will stop you, and it will stop you early.

Latenode fits a workflow with many steps that finishes quickly. The documentation says a ten-node scenario and a two-node one that both finish in a few seconds are each charged at least one CPU second, and that steps do not multiply the bill — the vendor says it itself rather than leaving you to infer it from an absent row. The three-minute run ceiling and the five active workflows are the limits to watch, and the unstated cost of idling is the gap you cannot plan around from the documentation.

Both free plans start without a card. The cheapest way to settle this is to put your actual workflow into both and read the counters — which is what we did, and why the two things that did not line up are in this article rather than left out of it.

FAQ

Is a CPU second comparable to a task?

No, and neither definition page offers a conversion. One measures how long a run took; the other counts how many times the run called an outside app. Ten thousand of one and a hundred of the other cannot be placed on the same scale, which is why this comparison is about which limit you hit first rather than which allowance is bigger.

Does building cost anything?

Latenode’s documentation covers one case and states it plainly: running a single node does not consume CPU seconds. That is one node, not a whole scenario executed as a test. Pabbly’s pages do not say, and the only reading we have is one workflow on October 25, 2026, where a test send consumed no task and the live run that followed consumed one.

Which free plan lets me build a longer workflow?

Latenode. Its free plan prints no limit on the number of steps, while Pabbly’s free card gives 2-Step Workflows. The trade is that Latenode’s free runs have to finish inside three minutes, while Pabbly’s pages say nothing about run length in either direction.

What happens to a workflow that mostly waits?

On Pabbly the waiting part is free: checks for new data count as operations, and operations are unlimited on every plan. The moment the workflow finds something and acts on an outside app, that action is a task like any other. On Latenode we cannot tell you what the waiting costs — the page that defines CPU seconds does not say what an idle check costs.

Why does my Latenode screen say credits instead of CPU seconds?

Older accounts still show the earlier unit. The documentation puts one credit at thirty CPU seconds, so a balance in credits and a balance in CPU seconds are not the same quantity even when the products behave identically.

Do either of them need a credit card to start?

No. Latenode’s free plan is stated as $0 a month forever with no card required, and Pabbly’s free card says the same. You can put the same workflow into both and compare the counters before paying anyone.

---

Sources & Verification

Hands-on readings reused here, with the dates they were taken: Latenode on September 26 and October 25, 2026, both on its free plan; Pabbly Connect on October 25 and October 29, 2026, both on its free plan. No new runs were made for this page.

Official sources (checked November 3, 2026):

• What Are CPU Seconds? (Latenode) — the definition, the one-second minimum, the node-count rule, the nested-scenario rule, the Pay as You Go scale and the legacy credit conversion
• Latenode pricing — the free plan’s 10,000 CPU seconds, five active workflows and three-minute run ceiling, each stated against its paid counterpart
• Pabbly Connect — the task definition, the unlimited-operations rule for polling checks, the named internal apps that are never counted, and the free card’s 2-Step Workflows row

Sign-up links for the two tools compared here: Latenode and Pabbly Connect. Both pay us, and both are reported above with the place where their own product and their own pages did not agree.