n8n, Make and Zapier each bill in a different unit: n8n counts workflow executions, Make counts credits, and Zapier counts tasks. The unit matters more than the monthly price, because it decides how the same automation adds up. This guide explains each count from the vendors' own pages, checked on October 2, 2026, and shows what an AI step does to the bill. GrossiWeb's workflow automation service works mainly in n8n and also works with Make and Zapier. Prices change often, so read every figure below as true on that date only.
n8n counts one run as one execution
n8n bills by execution, and its pricing page defines one as "a single run of your entire workflow." The number of steps and the amount of data do not change the count. On October 2, 2026 the page showed Starter at $20 a month billed annually ($24 on the monthly toggle) for 2.5K executions, and Pro at $50 billed annually ($60 monthly) for 10K executions. All plans list unlimited steps and users.
n8n's September 25, 2026 announcement of Agents says one agent turn is one execution, tool calls to your workflows are not counted separately, and agents share your execution quota. Agents are in preview, and the model bill goes to your own provider credentials or n8n's Gateway credits. The pricing page also notes that a standard self-hosted version is available on GitHub.
Make counts every module run as a credit
Make bills in credits. Its credits page says credits replaced operations as the billing unit, and for non-AI apps one operation equals one credit. Each module run is an operation, and a trigger counts once per check. Make's own example is a scenario whose trigger returns 10 form responses and then runs three more modules: 1 + (3 × 10) = 31 operations for that single run.
On October 2, 2026 the pricing page showed Free at $0 for up to 1,000 credits a month, and Core at $12 a month with annual billing ($16 with monthly billing) for 10k credits. Pro and Teams showed $21 and $38 on annual billing for the same 10k credits.
Make AI Agents are listed for all plans and labeled beta. With Make's own AI provider, credits depend on tokens and operations. With your own provider connection, such as OpenAI or Anthropic, you pay Make credits for operations and the provider for tokens. On August 25, 2026 Make split input and output token rates for its Small, Medium and Large models. Its example is a Medium step with 2,000 input tokens and 100 output tokens: about 0.6 credits before, about 0.16 after. The rate table confirms it: 2,000 ÷ 18,080 + 100 ÷ 2,260 ≈ 0.111 + 0.044 = 0.155.
Zapier counts every successful action as a task
Zapier bills in tasks, and its help center says a task is any successful action. Triggers, filters, paths, formatter steps and failed steps do not count. Zapier's own example is a Zap that uploads a file to Dropbox and posts in Slack: two tasks per run, so ten runs use 20 tasks.
On October 2, 2026 the pricing page showed Free at $0 with 100 tasks a month and two-step Zaps. Professional showed "Starting from" $19.99 a month with yearly billing, or $29.99 with monthly billing, for the smallest paid tier of 750 tasks, and the 1.5K tier showed $58.50 a month on monthly billing. At your limit, Zapier switches you to pay-per-task billing at a higher per-task rate unless you turn that off, in which case new runs are held until the next cycle.
AI steps can cost more tasks on Zapier. Its model tier page, updated October 1, 2026, lists task multipliers of 1x for Standard, 3x for Advanced and 5x for Premium, with Premium the default for new steps on paid plans and 1x for your own AI account. Each successful Zapier MCP tool call uses two tasks, and Zapier Agents use a separate activity quota (1,500 a month on the Agents Pro add-on).
A worked example using the vendors' own numbers
Assume a form sends 200 leads a month and each lead triggers four actions: create a CRM contact, email the lead, alert Slack and add a spreadsheet row. The 200 leads and four actions are assumptions; every price and counting rule comes from the pages above.
- n8n: 200 runs is 200 executions, against 2.5K included on Starter, however many steps the workflow has.
- Make: one trigger plus four modules is 5 operations per lead, so 5 × 200 = 1,000 credits against 10k on Core. This assumes the trigger runs once per lead. Make counts one operation per trigger check, so a scheduled trigger that checks often adds credits.
- Zapier: four successful actions is 4 tasks per lead because the trigger is free, so 4 × 200 = 800 tasks. That is 50 over the 750-task tier, so you either pay Zapier's higher pay-per-task rate for the extra 50 or move to the 1.5K tier at $58.50 a month on monthly billing.
Now add one AI step per lead to classify the inquiry, using Make's example size of 2,000 input and 100 output tokens. That size is an illustration, not a measurement.
- n8n: still 200 executions, with the model bill separate.
- Make: about 0.155 credits per lead for tokens on the Medium tier, so 0.155 × 200 = 31 credits, plus the credits the module itself uses.
- Zapier: a Standard step uses 1 task, so 4 + 1 = 5 tasks per lead, and 5 × 200 = 1,000 tasks. A Premium step uses 5 tasks, so 4 + 5 = 9 tasks per lead, and 9 × 200 = 1,800 tasks, over the 1.5K tier.
With your own Claude key, Anthropic's Sonnet 5.5 page lists $2 per million input tokens and $10 per million output tokens. One call costs 2,000 × $2 ÷ 1,000,000 = $0.004 plus 100 × $10 ÷ 1,000,000 = $0.001, or $0.005, so 200 calls cost $1.00. This guide does not rank the vendors from these lines, because currencies, billing terms and features differ, and your workflow will not match this one.
How to choose between them
Choose by the shape of your workflows, because each count rewards a different shape. A workflow with many steps or many items per run suits n8n's count, which does not grow with steps or data. Short jobs at low volume suit Zapier's entry tiers, and GrossiWeb's service copy calls Zapier the simplest option for a few straightforward two-step connections.
- If data must stay inside your own infrastructure, price self-hosted n8n first, which is one reason GrossiWeb uses it as its primary tool.
- If you plan heavy AI use, ask where the model bill lands: in the platform's credits or tasks, or with your own provider key.
- If you are considering an agent, start from a plain workflow and add an agent only where a step needs judgment. See custom AI agents.
What to do this week
- List your three busiest workflows, with the steps in each and the runs per month.
- Count each one three ways: runs for n8n, module runs for Make, and successful actions for Zapier, including any per-item loops.
- Mark every AI step, its model tier, and whether it uses your own key.
- Open each vendor's pricing page on the day you decide, and write the date beside every number.
To decide what to build first, read which AI tools waste time and which save it and the agent that saves 10 hours a week. GrossiWeb quotes automation builds after a scoping call and you pay the platform subscription directly. GrossiWeb's pricing page lists the prices it publishes, and where to start with AI covers the first step.