How Much Do AI Automations Really Cost? A Pricing Breakdown
The real ai automation cost is two bills added together: what the automation platform charges per step or per run, and what the AI model charges per token of text. For a small business running a few thousand AI steps a month, the AI part is often a few dollars, and the platform plan is usually the bigger number. This breakdown is for beginners and small teams who want to price a workflow before they build it.
Quick verdict
- Lowest entry price for many steps: Make, whose Core plan lists 10,000 credits a month for $9 billed monthly.
- Cheapest per long workflow: n8n, because it bills per whole run, not per step.
- Easiest but priciest per step: Zapier, which counts every successful action as a task.
- AI model cost: small "mini" or "Haiku" class models keep per-message costs well below a cent for short text.
All prices are as of October 2026, per each vendor's pricing page.
Methodology: what we checked
This is desk research, not a hands-on test. On 7 October 2026 we read the official pricing pages for Zapier, Make and n8n, the Zapier and Make help-center articles that explain how usage is counted, and the official API pricing pages for OpenAI and Anthropic. We then combined the two kinds of pricing in simple arithmetic.
Every worked example below is illustrative. It uses the cited prices and stated assumptions about message length, and your real numbers will differ.
How the AI automation cost adds up
Platform pricing units
| Platform | What you pay for | Entry paid plan (October 2026) | Free option |
|---|---|---|---|
| Zapier | Tasks: each successful action step | Professional from $19.99/mo billed annually, $29.99 monthly | 100 tasks/month, two-step Zaps |
| Make | Credits: most module actions use 1 credit | Core $9/mo billed monthly, 10,000 credits | 1,000 credits/month, 2 active scenarios |
| n8n Cloud | Executions: one full workflow run | Starter €20/mo billed annually, 2,500 executions | Self-hosted Community Edition |
Zapier
Zapier's help center says a task is any successful action. Triggers, filters, paths, Formatter, delays and failed actions do not count. AI by Zapier steps count as tasks. If you hit your limit without pay-per-task billing turned on, new runs are held until the cycle resets.
Make
Make bills in credits, which replaced operations one to one in August 2025. Each action module uses one credit per bundle it processes. When you connect your own OpenAI account, Make's help center says you pay Make credits for the operation and pay OpenAI separately for tokens. If you use Make's built-in AI Provider instead, credits are calculated from tokens, so one AI step can use more than one credit.
n8n
n8n counts one execution per workflow run, no matter how many nodes run inside it. Its Pro cloud plan lists 10,000 executions for €50 a month billed annually. The self-hosted Community Edition is free under n8n's Sustainable Use License, but you pay for hosting.
AI API pricing: OpenAI and Anthropic
AI providers charge per million tokens, separately for input (your prompt) and output (the reply). A token is roughly a short word or part of a word.
OpenAI API
| Model | Input per 1M tokens | Output per 1M tokens |
|---|---|---|
| gpt-5-nano | $0.05 | $0.40 |
| gpt-5-mini | $0.25 | $2.00 |
| gpt-4o-mini | $0.15 | $0.60 |
Source: OpenAI API pricing page, standard tier, as of October 2026. OpenAI lists many more models; larger ones cost considerably more.
Anthropic API
| Model | Input per 1M tokens | Output per 1M tokens |
|---|---|---|
| Claude Haiku 4.5 | $1 | $5 |
| Claude Sonnet 5.5 | $2 | $10 |
Source: Anthropic (Claude) pricing page, as of October 2026. The page also notes that batch processing cuts costs by 50%, which suits jobs that don't need an instant answer.
Worked examples: cost per workflow
Assumptions for these illustrative examples: each AI call sends about 1,000 input tokens (a typical email plus instructions) and gets back about 200 output tokens (a short summary or draft reply).
OpenAI API cost per email
With gpt-5-mini: 1,000 input tokens cost $0.00025 and 200 output tokens cost $0.0004. That is about $0.00065 per email, or roughly $0.65 per 1,000 emails.
With Claude Haiku 4.5: 1,000 input tokens cost $0.001 and 200 output tokens cost $0.001, about $0.002 per email, or roughly $2 per 1,000 emails.
Platform units for a four-step workflow
Workflow: new email arrives (trigger), AI summarises it, summary is added to a Google Sheet, a Slack message is sent.
| Monthly runs | Zapier tasks | Make credits (own OpenAI key) | n8n executions |
|---|---|---|---|
| 100 | 300 | about 400 | 100 |
| 1,000 | 3,000 | about 4,000 | 1,000 |
| 5,000 | 15,000 | about 20,000 | 5,000 |
At 1,000 runs, the AI part with gpt-5-mini is about $0.65 a month. The platform part fits Make Core's 10,000 credits ($9) or n8n Starter's 2,500 executions (€20 billed annually). On Zapier, 3,000 tasks sits above the entry Professional tier, so use the task slider on Zapier's pricing page to see the exact plan price for that volume.
At 5,000 runs, Make needs about 20,000 credits, which means a larger credit tier, while n8n Pro's 10,000 executions would still cover it. This is where per-run billing starts to pay off.
Which one to pick for your situation
- Under a few hundred runs a month: try the free tiers first. Your AI cost will be pennies.
- Simple workflows, you value ease: Zapier, and budget for tasks per action.
- Multi-step workflows on a tight budget: Make, and keep an eye on modules that loop over lists.
- Long workflows or high volume: n8n Cloud, or self-hosted if you have the skills.
For a broader comparison of features and ease of use, see our Zapier vs Make vs n8n guide. If you have settled on Make, our Make.com AI templates list estimates credits for seven common scenarios. Voice workflows come with their own separate pricing from the voice provider, which we cover in our guide to AI phone agents for small businesses.
Pricing and free-tier limits
Free tiers are useful for testing, not for running a business. Zapier's free plan only allows two-step Zaps, so an AI step plus an action already fills it. Make's free plan caps you at two active scenarios with a 15-minute minimum interval. n8n has no hosted free plan listed on its pricing page; the free route is self-hosting.
Hidden costs to budget for:
- Pay-per-task overages on Zapier if you enable them.
- Bundles multiplying credits on Make when a step handles lists.
- Server, backups and your time if you self-host n8n.
- Longer prompts and bigger models on the AI side.
Limitations
This article reflects vendor pricing pages on one date. Prices, tiers and currencies change, n8n lists prices in euros, and taxes vary by country. Token counts in the examples are assumptions; real emails vary a lot in length. We did not run these workflows or measure real bills. Treat the figures as a starting estimate and check your provider dashboards after the first week.
What none of them do well
None of the platforms show you the combined cost in one place. Zapier, Make and n8n report their own units, and OpenAI or Anthropic report token spend in a separate dashboard. You have to add the two yourself. Spending caps also differ, so set a usage limit with your AI provider before you switch on a high-volume workflow.
FAQ
How much does Zapier cost per month?
As of October 2026, Zapier's pricing page lists a Free plan with 100 tasks a month and a Professional plan from $19.99 a month billed annually ($29.99 monthly) at its lowest task tier. The price rises with the number of tasks you choose.
How much does the OpenAI API cost?
It depends on the model and how many tokens you send and receive. As of October 2026, OpenAI lists gpt-5-mini at $0.25 per million input tokens and $2.00 per million output tokens, and larger models cost more.
Is self-hosting n8n cheaper?
The Community Edition software is free, so your cost is the server plus your own time for setup, updates and backups. For someone comfortable running a server it can be cheaper at high volume; for a beginner, n8n Cloud or another hosted tool is usually the safer choice.
Conclusion
Before you build, write down three numbers: steps per run, runs per month and rough text length. Multiply them against the units in this article, pick the cheapest small model that does the job, and set a spending limit with your AI provider. That gives you a realistic monthly figure in ten minutes.
Sources and methodology
Desk research of official pages, last reviewed 7 October 2026. No hands-on tests.
- Zapier pricing: https://zapier.com/pricing
- Zapier help center, task usage: https://help.zapier.com/hc/en-us/articles/8496196837261
- Make pricing: https://www.make.com/en/pricing
- Make help center, credits: https://help.make.com/credits
- n8n pricing: https://n8n.io/pricing/
- OpenAI API pricing: https://developers.openai.com/api/docs/pricing
- Anthropic (Claude) pricing: https://claude.com/pricing