Choose n8n for long or high-volume workflows, code, AI agents and self-hosting: it bills one execution per workflow run, and its Community Edition is free to self-host under the Sustainable Use License. Choose Zapier when non-technical teammates build the automations or you need a niche app: it lists 9,000+ apps and bills a task for each successful action. Choose Make for visual, branching scenarios at the lowest entry price: it bills credits, usually one per module action, from $9 a month billed annually for 5,000 credits. In our hypothetical examples, Make was cheapest for short and mid-sized workflows, n8n Cloud for a ten-step workflow at 5,000 runs a month, and Zapier cost the most every time.
Prices, limits and features checked on each vendor's official pages on October 1, 2026. This update replaces the September 2026 version, which gave no prices, and absorbs our separate n8n vs Make vs Zapier article, whose cost table treated Zapier tasks and Make operations as n8n executions and called n8n open source. We teach n8n in our automation course, so weigh our view accordingly; every price links to its source.
The short answer
- Choose n8n for workflows with many steps or high volume, JavaScript or Python code, AI agents with approval steps, and when data must stay on your own servers.
- Choose Zapier when non-technical teammates build the automations, when you need a ready-made connector for a niche app, or when you want nothing to maintain.
- Choose Make for visual scenarios with routers and branches at the lowest entry price, without running servers.
n8n vs Zapier vs Make at a glance
| n8n | Zapier | Make | |
|---|---|---|---|
| Billing unit | Execution: one per full workflow run | Task: one per successful action | Credit: most module actions use one |
| Free option | Self-hosted Community Edition; free Cloud trial | 100 tasks a month, two-step Zaps | 1,000 credits a month, 2 active scenarios |
| Entry paid plan, billed annually | Starter: 20 euros a month for 2,500 executions | Professional: $19.99 a month for 750 tasks | Make plan: $9 a month for 5,000 credits |
| Integrations (vendor's own count) | Over 500, plus HTTP Request for any REST API | 9,000+ apps | 3,000+ apps |
| Custom code | JavaScript and Python (Code node) | JavaScript and Python (Code by Zapier) | JavaScript and Python (Make Code app) |
| Self-hosting | Yes | No | No (an on-prem agent can reach local apps) |
| Where hosted data lives | Frankfurt, Germany (n8n Cloud) | AWS in the United States | AWS in the EU or North America |
| License | Sustainable Use License (source-available) | Closed-source cloud service | Closed-source cloud service |
Sources: n8n pricing, the n8n homepage, Zapier pricing, Code by Zapier, Zapier security and Make pricing. n8n shows prices in euros; Zapier and Make in US dollars.
How each one bills: executions, tasks and credits
The headline prices hide three different meters, and the meter matters more than the price:
- n8n counts executions. Its pricing page says: "An execution is a single run of your entire workflow. It doesn't matter how many steps are in the workflow or how much data it processes." Manual test runs, sub-workflow runs and error workflow runs don't count toward a paid plan's quota, per n8n's executions docs.
- Zapier counts tasks. "A task is any successful action that runs in Zapier," per its task usage guide. Zapier does not bill trigger steps, Filter and Paths steps, failed steps, or built-in tools such as Formatter, Delay and Looping. AI by Zapier steps use a number of tasks that depends on the model, Code by Zapier uses one task per run (more with extended runtime), Zapier MCP uses 2 tasks per successful tool call, and steps inside error-handler paths count.
- Make counts credits. Its pricing page says each module action, "like adding a Google Sheet row or fetching Gmail account data, counts as one credit." That includes reading data from an app or webhook, searching, transforming and iterating, while routers and error-handler modules use none. The Make Code app uses 2 credits per second of run time, and Make's own AI features can use more than one credit (credit docs).
Plans and prices in October 2026
n8n
| Plan | Price per month | Executions a month | Notable features |
|---|---|---|---|
| Starter | 20 euros billed annually (24 euros monthly) | 2,500 | 5 concurrent executions |
| Pro | 50 euros billed annually (60 euros monthly) | 10,000 (larger tiers available) | Up to 50 concurrent executions, admin roles, workflow history |
| Business (self-hosted) | 667 euros billed annually (800 euros monthly) | 40,000 | SSO, Git version control, environments |
| Enterprise | Custom | Custom | 200+ concurrent executions, log streaming, SLA support |
Every plan includes unlimited users and workflows and every integration (n8n pricing). The self-hosted Community Edition has no license fee.
Zapier
| Tasks a month | Professional, billed annually | Professional, monthly | Team, billed annually |
|---|---|---|---|
| 750 | $19.99 | $29.99 | Not offered |
| 1,500 | $39 | $58.50 | Not offered |
| 2,000 | $49 | $73.50 | $69 |
| 5,000 | $89 | $133.50 | $119 |
| 10,000 | $129 | $193.50 | $169 |
| 20,000 | $189 | $283.50 | $249 |
| 50,000 | $289 | $433.50 | $399 |
Monthly prices from the task selector on Zapier's pricing page. The Free plan includes 100 tasks a month with two-step Zaps. Professional adds multi-step Zaps and premium apps; Team adds 25 users and shared Zaps and folders.
Make
New sign-ups now see a single paid plan, the Make plan, priced by monthly credits, next to a Free plan and a custom Enterprise plan. Older guides, and parts of Make's own FAQ, still mention Core, Pro and Teams plans.
| Credits a month | Make plan, billed annually | Make plan, monthly |
|---|---|---|
| 5,000 | $9 | $10.59 |
| 10,000 | $16 | $18.82 |
| 20,000 | $29 | $34.12 |
| 40,000 | $53 | $62.35 |
| 80,000 | $91 | $107.06 |
Monthly prices from the credit slider on Make's pricing page. The Free plan includes 1,000 credits a month. Extra credits come in bundles of 1,000 or 10,000.
Worked cost example (hypothetical)
Everything in this section is hypothetical. It shows how the three meters play out, not what your workflows will cost. Both workflows start from an instant webhook, so no polling is involved:
- Lead intake: a trigger plus 3 actions (create a CRM contact, post to Slack, add a sheet row). Each run uses 1 n8n execution, 3 Zapier tasks and 4 Make credits.
- Order processing: a trigger plus 9 actions. Each run uses 1 execution, 9 tasks and 10 credits.
| Workflow and volume | n8n Cloud | Zapier Professional | Make plan |
|---|---|---|---|
| Lead intake, 1,000 runs a month | 1,000 executions: Starter, 20 euros | 3,000 tasks: 5,000-task tier, $89 | 4,000 credits: 5,000-credit step, $9 |
| Lead intake, 5,000 runs a month | 5,000 executions: Pro, 50 euros | 15,000 tasks: 20,000-task tier, $189 | 20,000 credits: 20,000-credit step, $29 |
| Order processing, 1,000 runs a month | 1,000 executions: Starter, 20 euros | 9,000 tasks: 10,000-task tier, $129 | 10,000 credits: 10,000-credit step, $16 |
| Order processing, 5,000 runs a month | 5,000 executions: Pro, 50 euros | 45,000 tasks: 50,000-task tier, $289 | 50,000 credits: 80,000-credit step, $91 |
Monthly prices on annual billing, using the smallest plan that covers each volume. n8n's prices are in euros and the others in US dollars; we have not converted them. Self-hosted n8n would show no license fee in every row, but you would pay for a server and the time to run it.
What it shows: Make has the lowest price in three of the four cases. n8n charges the same whether a run has 4 steps or 10, so it pulls ahead once workflows are long and busy. Zapier costs the most in every case because each action is a task; in return you get the largest app catalog and the simplest editor. Real workflows add AI steps, retries, error paths and polling triggers, so price your own on each pricing page, and use our workflow automation ROI method to weigh the subscription against build and maintenance time.
Limits that catch people out
- Run time: n8n Cloud Starter stops an execution after 5 minutes and Pro after 40. Make allows 5-minute runs on Free and 40 minutes on the Make plan.
- Polling and scheduling: Zapier checks for new data every 15 minutes on Free, every 2 on Professional and every minute on Team. Make's minimum interval between scheduled runs is 15 minutes on Free and 1 minute on the Make plan.
- Concurrency: n8n Cloud runs 5 executions at once on Starter and up to 50 on Pro.
- Hitting the cap: Zapier switches to pay-per-task billing at a higher per-task rate, or holds tasks until the next cycle if you turn that off, and holds everything once usage reaches three times your plan's limit. Make stops scenarios until you add credits, queuing incoming webhooks up to your plan's allowance. n8n's Business plan bills overages at 4,000 euros per extra 300,000 executions.
- History: n8n keeps execution logs for 7 days on Starter and 30 on Pro; Make keeps them for 7 days on Free and 30 on the Make plan.
Integrations: how many apps each one connects
Each vendor counts differently, so treat these as rough guides: Zapier lists 9,000+ apps, Make 3,000+ apps and n8n over 500 integrations. Whether the specific triggers and actions you need exist matters more than the total. All three can reach apps without a ready-made connector: n8n through its HTTP Request node, and custom nodes when self-hosted; Make through custom apps that connect to "virtually any application or service that provides a public API"; and Zapier through Webhooks by Zapier and Code by Zapier steps.
Self-hosting and licensing
Only n8n can be self-hosted. Zapier and Make are cloud services; Make offers an on-prem agent so scenarios can reach apps inside your network.
n8n's Community Edition is free to self-host under the Sustainable Use License, which lets you "use or modify the software only for your own internal business purposes or for non-commercial or personal use." Because open source licenses can't restrict use, n8n says "we do not call ourselves open source" and describes its model as fair-code. In practice, per n8n's license FAQ:
- You can run your own business on it and build automations for clients on your instance, as long as the clients can't create or edit them, and charge for setup and maintenance.
- You can't host n8n as a service where clients build their own workflows, or white-label it, without a commercial agreement.
- Features such as SSO, Git version control and environments need a paid license key, such as the self-hosted Business plan.
Self-hosting moves the work to you: a server, HTTPS, backups, updates and security. n8n's own server guide "recommends self-hosting for expert users," and n8n 3.0, scheduled for October 2026, supports Docker installs only, per its breaking-changes page. Our n8n self-hosting guide walks through a production setup.
AI and agent features
| Feature | n8n | Zapier | Make |
|---|---|---|---|
| Agents | AI Agent node on every plan | Zapier Agents: 400 activities a month free; Pro $33.33 a month billed annually for 1,500 | Make AI Agents (beta) on all plans |
| MCP | MCP Server Trigger and MCP Client Tool nodes, plus a built-in MCP server | Zapier MCP, at 2 tasks per successful tool call | Make MCP server |
| AI model costs | Paid to the model provider you connect | AI by Zapier steps use tasks that vary by model | Make's AI Provider charges credits by tokens; with your own key, the provider bills tokens |
Sources: n8n's MCP Server Trigger docs, Zapier pricing, Make pricing and Make's credit docs. n8n also has human review for tools, which pauses an agent until a person approves or denies a specific tool call, such as sending a message or deleting a record. That matters because OWASP ranks prompt injection as the top risk for LLM applications and recommends human approval for privileged operations.
Which one should you pick?
Pick Zapier if
- The people building automations are not technical.
- You rely on niche SaaS apps that only Zapier connects to out of the box.
- Most workflows are short, with one or two actions, or run rarely.
Pick Make if
- You want visual scenarios with routers, iterators and branches without writing code.
- You want the lowest entry price for low to moderate volume.
- Cloud hosting in the EU or North America suits you.
Pick n8n if
- Workflows have many steps or run thousands of times a month.
- You write JavaScript or Python, or call APIs that have no connector.
- You are building AI agents and want approval steps before risky actions.
- Data must stay on your own servers, or you build automations for clients within the license.
Switching later is a rebuild, not an import. Our n8n vs Zapier guide maps Zapier steps to n8n nodes and covers a safe migration, and our Make vs n8n comparison looks at those two side by side. For head-to-head summaries, see n8n vs Zapier and n8n vs Make.
Next steps
Before you buy, list your five busiest workflows with their steps and monthly runs, price them on all three pricing pages, and run the result through our ROI calculator. If you are still deciding what to automate, start with our business automation tools guide. If you choose n8n, set up error handling before anything goes live. If you would rather build your own AI product than automate other tools, AI SaaS Builder covers validating an idea, building it on Supabase, Next.js and the Claude API, and charging with Stripe.
Want the full AI SaaS Builder playbook?
A 10-module, 52-lesson curriculum: validate an idea, build on Supabase and Next.js, add AI features with the Claude API, speed up with Claude Code and MCP, deploy on Vercel, launch, and charge with Stripe.
n8n vs Zapier vs Make FAQ
Which is cheaper: n8n, Zapier or Make?
It depends on how many steps each run has and how often it runs. On October 1, 2026, the entry paid plans billed annually were Make at $9 a month for 5,000 credits, Zapier Professional at $19.99 a month for 750 tasks and n8n Cloud Starter at 20 euros a month for 2,500 executions. In our hypothetical examples, Make was cheapest for short and mid-sized workflows, n8n Cloud for a ten-step workflow running 5,000 times a month, and Zapier cost the most every time. Self-hosted n8n has no license fee, but you pay for a server and the time to run it.
Can I self-host n8n, Zapier or Make?
Only n8n. Its Community Edition is free to self-host under the Sustainable Use License, and paid self-hosted plans add features such as SSO and Git version control. Zapier is hosted on AWS in the United States and Make on AWS in the EU or North America; neither has a self-hosted edition, though Make offers an on-prem agent that can reach apps inside your network.
Is n8n open source?
Not in the OSI sense. n8n's source code is public, but its Sustainable Use License allows use only for your own internal business purposes or for non-commercial or personal use, so n8n calls its model fair-code rather than open source. You can build automations for clients on your own instance as long as they can't create or edit them; hosting n8n as a service where clients build their own workflows needs a commercial agreement.
Which has the best AI agent features?
All three now offer agents and MCP. n8n gives the most control: an AI Agent node on every plan, MCP server and client nodes, and human approval for specific tool calls, with tokens billed by your own model provider. Zapier Agents is the quickest start, with 400 free activities a month and a Pro plan at $33.33 a month billed annually for 1,500. Make AI Agents is in beta on all plans and charges credits by tokens when you use Make's AI Provider.
Is n8n hard to learn?
It takes longer to learn than Zapier. n8n is visual, but it rewards comfort with APIs, JSON and expressions, plus optional JavaScript or Python. Zapier's step-by-step editor is the easiest for non-technical users, and Make sits between the two. Self-hosting adds server work on top: n8n itself recommends self-hosting for expert users, so start on n8n Cloud if you are new to it.
Can I migrate from Zapier to n8n or Make?
Yes, but plan to rebuild rather than import: n8n's documentation does not describe a Zapier importer. List each Zap's trigger, actions, filters and code steps, rebuild one low-risk workflow first, run old and new side by side with writes going to test destinations, then switch over and keep the old version until you trust the new one.