Make vs n8n: Visual Automation vs Open-Source Workflows In 2027
Dominic Tarn
Head Writer - ReVerb
11.02.2026
Make vs n8n compared: credit-based cloud pricing vs per-execution plans, free self-hosting, AI agents, code steps, data control and which tool fits your team.
If you have outgrown simple two-step automations, or you are wincing at a Zapier bill, the shortlist usually narrows to two names: Make and n8n. Both offer a visual canvas for multi-step workflows, AI agent building blocks and JavaScript or Python code steps.
The buyers are different, though. Make tends to land with operations, marketing and RevOps teams who want a polished hosted product and a huge app catalog. n8n tends to land with technical teams, agencies and AI builders who want to own the infrastructure, write code freely and stop paying per step.
The one-sentence difference: Make is a fully hosted visual automation platform that bills for every action a scenario performs, while n8n is a source-available workflow engine that you can self-host for free and that bills its cloud customers per full workflow run, no matter how many steps the run contains.
Make vs n8n At A Glance
Factor
Make
n8n
Deployment
Cloud only (US and EU data zones)
n8n Cloud, or self-hosted on your own servers
Billing unit
Credits: most module actions use 1 credit
Executions: one full workflow run, any number of steps
Free option
Free plan with 1,000 credits per month
Self-hosted Community Edition (no license fee); cloud trial only
Entry paid plan
Core from $9/month billed annually (10,000 credits)
Starter at €20/month billed annually (2,500 executions)
Mid-tier plan
Pro from $16/month annually; Teams from $29/month annually
Pro at €50/month annually (10,000 executions)
App catalog
3,000+ apps (directory lists about 3,600)
2,000+ integrations (directory lists about 2,600)
Custom code
Make Code app: JavaScript and Python, paid plans, 2 credits per second of run time
Code node: JavaScript and Python on every plan; npm modules when self-hosted
AI agents
Make AI Agents, built on the scenario canvas, with a reasoning panel
AI Agent node with chat model, tools, memory and MCP connections
Running out of quota
Scenarios stop until you add credits (auto-purchase available)
Workflows keep running; overage charges may apply (capped on Starter and Pro)
Best for
Non-developers and ops teams who want hosted, visual, app-heavy automation
Technical teams that want self-hosting, heavy code use or high-step workflows
The Core Difference: Metered Actions vs Owned Infrastructure
Make sells a managed product. You build scenarios on its canvas, Make runs them on its servers, and every action those scenarios take draws down a monthly pool of credits. Make’s help center describes credits as “the currency you buy and consume to use Make.” For ordinary apps one operation equals one credit, so a scenario that reads a form, looks up a CRM record, updates a sheet and sends a Slack message costs about four credits per run. Some AI and code features cost more, priced by tokens or run time.
n8n sells a workflow engine, and gives you a choice about who runs it. On n8n Cloud, the pricing page states that “an execution is a single run of your entire workflow,” so a 4-step workflow and a 40-step workflow cost the same. Or you can skip n8n’s billing entirely and run the self-hosted Community Edition from GitHub on your own server, where you pay for hosting instead of for usage.
That single difference shapes everything else. Make rewards short, simple scenarios and charges more as workflows get longer or loop over many items. n8n rewards long, complex workflows and data-heavy loops, but asks more of you technically, especially if you self-host.
Pricing Compared
Make Pricing
Make has five plans: Free, Core, Pro, Teams and Enterprise. The paid plans are priced by credit volume, so each plan has a slider rather than one fixed price. The figures below come from the USD price data that feeds Make’s pricing page, checked in November 2026:
Free: $0, 1,000 credits per month, 1 MB of data storage and one data store.
Core: $9/month billed annually or $10.59 billed monthly at 10,000 credits. 20,000 credits costs $16 annually ($18.82 monthly), 40,000 costs $29 ($34.12), and the slider tops out at 300,000 credits for $182.16 ($214.31).
Pro: $16/month annually or $18.82 monthly at 10,000 credits, $29 ($34.12) at 20,000 and $53 ($62.35) at 40,000. Pro scales to 8 million credits.
Teams: $29/month annually or $34.12 monthly at 10,000 credits, $53 ($62.35) at 20,000 and $99 ($116.47) at 40,000.
Enterprise: Custom pricing through sales, with wire transfer after invoice available.
Annual prepayment saves roughly 15% over monthly billing. Each 10,000 credits also comes with a usage allowance: 5 GB of data transfer, 10 MB of data storage, 10 MB of incomplete-execution storage and a webhook queue of 667 items.
What drives the bill up: long scenarios, iterators that loop over many records, high-frequency polling, Make’s built-in AI provider (billed by tokens on top of operations) and the Make Code app (2 credits per second of execution). Unused credits expire at the end of the term. If you run out, Make’s FAQ is direct: “Your scenarios won’t continue to run, until credits are added again.” You can buy 1,000 or 10,000 credit bundles, or turn on auto-purchase on Core, Pro and Teams.
Make’s pricing page does not publish per-plan feature limits (such as minimum scheduling interval) in readable form, so confirm those in the plan cards before you commit.
n8n Pricing
n8n’s pricing page lists prices in euros, as shown on the vendor’s site, and only displays annual billing rates for its paid plans. Every plan includes unlimited users, unlimited active workflows and all integrations.
Community Edition: Self-hosted from GitHub with no license fee. You pay for your own server, database and maintenance time.
Starter: €20/month billed annually, hosted by n8n. 2,500 executions per month, 5 concurrent executions, 1 shared project, forum support.
Pro: €50/month billed annually, hosted by n8n. 10,000 executions per month, up to 50 concurrent executions, 3 shared projects, admin roles and longer maximum execution duration.
Business: €667/month billed annually, currently self-hosted only. 40,000 executions, 6 shared projects, SSO and LDAP, Git version control and queue mode. n8n positions it for companies with fewer than 100 employees, and its Startup plan gives qualifying companies under 20 employees 50% off Business.
Enterprise: Contact sales. Custom execution volume, 200+ concurrent executions, cloud or self-hosted, dedicated support with an SLA and log streaming.
Cloud trials include Pro-level features and 1,000 executions with no credit card. When you exceed your quota, n8n says “your workflows will continue running without interruption, but overage charges may apply,” with overages capped on Starter and Pro. n8n Cloud stores data in the EU (Frankfurt).
What drives the n8n bill up: the number of times workflows run, not their length. A webhook that fires 50,000 times a month will push you into Business or Enterprise territory even if each run is a single step. On self-hosted Community Edition, the real costs are server size, database, backups, upgrades and the engineer who looks after them.
What You Actually Pay
Take a five-person marketing operations team running two kinds of automation. The first is lead routing: a form submission triggers a 6-step workflow (enrich, check CRM, create or update the contact, assign an owner, notify Slack, log to a sheet) about 1,500 times a month. The second is a nightly sync that pulls 200 records and runs each through 5 steps, every night.
On Make, lead routing uses about 9,000 credits (1,500 runs x 6 actions). The nightly sync loops 200 records through 5 modules, roughly 1,000 credits a night or about 30,000 a month. Total: around 39,000 credits, which fits the 40,000-credit tier. That is $29/month on Core billed annually ($34.12 monthly), or $53/month on Pro if you need Pro-level features. Teams costs $99/month annually at that volume.
On n8n Cloud, the same work is about 1,530 executions a month (1,500 lead runs plus 30 nightly syncs, since each sync is one run no matter how many records it loops through). That fits Starter at €20/month billed annually. Self-hosted Community Edition would cost only the server, typically a small virtual machine, plus the time to maintain it.
Flip the scenario and the math changes. A 2-step webhook running 8,000 times a month uses about 16,000 credits on Make (Core at $16/month annually) but needs n8n Pro at €50/month. Short, high-frequency workflows favor Make; long, loop-heavy ones favor n8n.
Ease Of Use And The Building Experience
Make
Make’s canvas is its signature. Modules appear as circles connected left to right, routers split paths visually, and you can watch data bubble through each module when you run a scenario. Field mapping is point-and-click, and built-in functions handle most transformations without code. Routers and error handlers do not consume credits.
For a marketer or operations lead with no programming background, Make is usually the faster tool to become productive in. The tradeoff: complex data manipulation has its own learning curve, and every extra module costs credits.
n8n
n8n’s editor is a node graph that will feel familiar to developers. Data passes between nodes as JSON items, and expressions use JavaScript syntax. That is flexible, but non-technical users face a steeper learning curve.
If your team already writes code, n8n often feels faster than Make because you can solve a transformation in five lines of JavaScript instead of three chained modules. If nobody on the team codes, Make is the gentler start.
AI Agents And Code Steps
Make
Make AI Agents are built on the same canvas as scenarios. You configure an agent with a model, instructions and inputs, give it access to Make’s app integrations as tools, and reuse it across multiple scenarios. A Reasoning panel shows each decision the agent makes, and you can add rules and manual approvals. Make frames agents as working “alongside deterministic logic, not instead of it.”
Billing for agents depends on the model connection. With Make’s own AI provider (available on all plans), an agent run costs 1 credit per operation plus token-based credits. With your own OpenAI or Anthropic connection (paid plans only), you pay 1 credit per operation and your provider bills the tokens directly. Make also runs an MCP server that lets clients such as Claude and ChatGPT run your scenarios as tools; scenario run scopes work on all plans, while management scopes need a paid plan.
For code, the Make Code app runs arbitrary JavaScript or Python inside a scenario. It is available on Core, Pro, Teams and Enterprise, costs 2 credits per second of execution, and on Core through Teams runs with 1 CPU, 512 MB of RAM and a 30-second limit. Enterprise gets 2 CPUs, 1 GB of RAM, 300 seconds and custom library imports. Other plans are limited to a preset list of libraries (lodash and moment for JavaScript, requests, pendulum and toolz for Python).
n8n
n8n has invested heavily in AI and it shows. The AI Agent node connects to a chat model and one or more tool sub-nodes, and the agent decides which tools to call. You can attach memory, vector stores and multiple LLM providers (OpenAI, Anthropic and Google among them) in one workflow, connect an agent to an MCP registry server in one click, and use evaluations to test AI workflows. Pro adds metric-based evaluations. An agent run that calls ten tools still counts as one execution on n8n Cloud; model tokens are billed by your provider.
The Code node supports JavaScript and Python on every plan, with no per-second charge. Native Python runs through task runners and is stable as of n8n 2. Self-hosted users can import built-in and external npm modules once enabled, while n8n Cloud limits imports to the crypto module and moment. If your workflows lean on custom libraries, self-hosted n8n is the most permissive option in this comparison.
Integrations And App Catalog
Make claims more than 3,000 apps, and its integrations directory listed about 3,600 when we checked. It also offers enterprise connectors such as Workday and ServiceNow, plus a custom app builder for any public API.
n8n claims more than 2,000 integrations, and its directory showed about 2,600. The core catalog is smaller and some nodes cover fewer actions than Make’s equivalents, but n8n’s HTTP Request node and code steps make it easy to call any API directly, and every plan includes all integrations. Check both directories for the exact actions your stack needs.
Make is cloud only. Organizations are assigned to a data zone: US1, US2, EU1 or EU2. That gives European teams an EU option, but Make’s own documentation warns that choosing an EU zone “does not automatically mean every request gets processed and stays in the EU.” Several AI features, including Make AI Web Search and all OpenAI models through Make’s AI provider, are processed in the US regardless of zone. If your compliance team needs strict residency, review Make’s AI providers and data zones page model by model.
n8n
n8n gives you three choices: n8n Cloud (data stored in Frankfurt), self-hosted on your own infrastructure, or air-gapped deployments for Enterprise. Self-hosting means your workflow data, credentials and execution logs sit on servers you control, though any external app or AI model a workflow calls still receives the data you send it. n8n lists GDPR and SOC 2 Type 2 compliance on its enterprise page.
Licensing deserves a careful read. n8n is not open source in the OSI sense. It uses the Sustainable Use License, which says: “You may use or modify the software only for your own internal business purposes or for non-commercial or personal use.” In practice, n8n’s license FAQ confirms you can use it free inside your company, and agencies can charge clients for building and maintaining workflows, provided clients do not access your n8n instance directly. You cannot host n8n as a service where your customers build their own workflows, or white-label it, without a commercial agreement. Files marked “.ee” (enterprise features) need an Enterprise license.
The free Community Edition also leaves out several team features: SSO, Git-based version control, environments, external secrets, log streaming, projects, custom variables and sharing of workflows and credentials. Registering for a free community license key unlocks folders, debug in editor and custom execution data. Larger teams that need SSO and Git self-hosted will be looking at the Business plan.
Scale, Limits And Reliability
On Make, scale is a credit question. Pro and Teams sliders run to 8 million credits per month, and Make handles infrastructure, queues and uptime. The main operational risk is the hard stop: when credits run out, scenarios pause until you top up, while incoming webhooks queue up to your plan’s limit. Auto-purchase of extra credits removes most of that risk for paid plans.
On n8n Cloud, scale is about executions and concurrency. Starter allows 5 concurrent executions and caps execution duration at 5 minutes, Pro raises concurrency to as many as 50 and duration to 40 minutes, and Enterprise starts at 200 or more concurrent executions. Self-hosted, there is no execution meter, but you are responsible for scaling. Queue mode, which spreads work across multiple workers, is listed as a Business feature, and multi-main mode for high availability is not included in Community Edition.
Support is worth checking before you buy. Make does not spell out support tiers on its pricing page, so confirm what your plan includes. n8n’s Starter and Business plans list forum support, and dedicated support with an SLA is an Enterprise feature. Self-hosted Community Edition users rely on the community forum and their own team.
Which Should You Choose
Choose Make If…
Your automation builders are marketers, ops staff or analysts rather than developers.
You want a fully hosted product with no servers, upgrades or backups to manage.
Most of your workflows are short (a handful of steps) and you value deep, ready-made app modules.
You want a low-cost entry point: $9/month billed annually for 10,000 credits is hard to beat for light use.
Choose n8n If…
You have developers on the team and expect to write JavaScript or Python regularly.
Your workflows are long, loop over large datasets or chain many AI tool calls, where per-step billing would get expensive.
You need data to stay on your own infrastructure, or you want to avoid usage fees entirely by self-hosting Community Edition.
You are building AI agents and want fine control over models, memory, vector stores and MCP connections.
If neither fits cleanly, our guide to the top n8n alternatives covers other self-hostable and hosted options.
Frequently Asked Questions
Is n8n Really Free?
The self-hosted Community Edition has no license fee and is available on GitHub. You still pay for hosting and maintenance, and the Sustainable Use License restricts use to internal business, personal or non-commercial purposes. n8n Cloud has no free plan, only a free trial with 1,000 executions.
Is Make Cheaper Than n8n?
For short, simple workflows at low volume, usually yes: Make Core starts at $9/month billed annually for 10,000 credits, while n8n’s cheapest cloud plan is €20/month for 2,500 executions. For long or loop-heavy workflows, n8n Cloud is often cheaper because it counts one execution per run instead of one credit per step. Self-hosted n8n is cheapest at scale if you have the skills to run it.
What Is The Difference Between Make Credits And n8n Executions?
A Make credit is consumed by each action a module performs, so one scenario run can use anywhere from 2 credits to thousands. An n8n execution is a single run of a whole workflow, regardless of how many steps or items it processes. AI token usage and code run time add extra credits on Make, while n8n does not meter them per step.
Can An Agency Use n8n To Build Workflows For Clients?
Yes. n8n’s license FAQ says consulting and automation services are explicitly permitted, and you can charge for workflow setup and maintenance. The limit is that clients should not access your n8n instance directly. Hosting n8n so customers build their own workflows, or white-labeling it, requires a commercial agreement.
Which Is Better For AI Agents, Make or n8n?
Both are capable. Make AI Agents suit teams that want agents inside visual scenarios with a reasoning panel and approval steps, without code. n8n’s AI Agent node offers more control over models, memory, vector stores and MCP tools, and its per-execution billing makes multi-step agent runs more predictable on cost.
Conclusion
Make and n8n solve the same problem from opposite ends. Make is the more approachable, fully managed platform, with a larger catalog of deep app modules and a cheap entry tier, but its per-action credits add up quickly on long or looping workflows. n8n asks more of its users, and its license is fair-code rather than open source, but it rewards complex workflows with per-run billing and gives you the option of running everything on your own servers.
The deciding questions are simple: who will build the workflows, how many steps each one runs, and where your data needs to live. If your builders are non-technical and your workflows are short, start with Make. If you have developers, heavy AI or data workloads, or strict data control requirements, start with n8n. Both offer a free way in, so build your two most important workflows on each and compare the real usage numbers before you commit.
