What Is n8n? How It Works, Pricing, AI Agents (2026)

What is n8n? n8n is a workflow automation tool. You connect apps and services as nodes on a canvas, and when one thing happens, such as a form being filled or a timer firing, the next steps run on their own. It adds code and AI agents where drag-and-drop runs out. You can run it free on your own machine, or pay n8n to host it from €20 a month.

This guide was checked on 30 October 2026 from Delhi. We read n8n's pricing page, documentation, license, release notes and GitHub repository that day, plus the funding reports Wikipedia cites. Below: what n8n is and how it works, what it is used for, whether it is free, what it costs a month, what the n8n AI agent is used for, whether it is open source, and what the 3.0 release changes.

What is n8n?

n8n is a workflow automation platform. A workflow is a chain of steps that runs when something triggers it. Each step is a node, a box on the canvas that stands for an app, a service or an operation. You connect the boxes, tell each one what to do, and n8n passes data from one to the next. Where the boxes fall short, you write JavaScript or Python in a Code node, or hand a decision to an AI agent.

The name is a numeronym: "nodemation", from the node view, the Node.js runtime and "automation", squeezed to an n, eight letters and an n. Jan Oberhauser, who had worked as a visual-effects pipeline engineer, launched it in Berlin in October 2019. That history is in Wikipedia's article on n8n, which cites TechCrunch's report on the first funding round.

n8n.io on 30 October 2026 from Delhi.

On 30 October 2026 the n8n.io homepage led with "Build the agent. Understand its workflow." above a sample refund workflow: a request comes in, Postgres looks up the order, Slack asks for approval, and an If node sends it to a refund or a Zendesk ticket. That one picture is most of what n8n is.

n8n describes itself, in the llms.txt file it publishes for AI tools, as a platform that "combines AI capabilities with business process automation, giving technical teams the flexibility of code with the speed of no-code". The GitHub repository is plainer: "Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations." The repository showed 209,117 stars and 61,317 forks that day.

It sits in the same category as Zapier and Make, with two differences that explain most of its growth. You can run it yourself, on a laptop, a small virtual server or a company's own cluster. And the source code is public. Technical teams pick it for those reasons; the trade-off is that someone has to look after the install.

How does n8n work?

n8n works in three parts: a trigger starts a workflow, nodes do the work in order, and data moves between them as JSON. Every production workflow needs at least one trigger node, which n8n's glossary says is there "to determine when the workflow should run". Common triggers are:

• a schedule (every hour, every Monday at 9);

• a webhook, an address that another system can call;

• a chat message, for agent workflows;

• an app event: a new row in Google Sheets, a new email in Gmail, a Stripe payment, a form submission.

After the trigger, each node receives the output of the one before it. That output is a list of items, each a JSON object, and you reach into it with expressions. An expression is a small piece of JavaScript inside double curly braces, so a Slack node can post {{ $json.customer_name }} from the row a Google Sheets node just read. If JSON or webhooks are new to you, our explainer on what an API is covers the vocabulary n8n assumes.

The Code node is where n8n leaves the no-code lane. It runs JavaScript, or Python, over the items passing through. A transformation that takes six drag-and-drop steps in another tool is ten lines here, and a pattern from our regex cheat sheet drops straight into a Code node. The HTTP Request node calls any API that has no dedicated node. Together they are why developers call n8n low-code rather than no-code: the canvas handles the routine, the code handles the rest.

Executions are logged. Each run shows which node received what, which one failed and with what error, and a failed run can be retried from the node that broke. Workflow history keeps earlier versions for one day on Starter, five on Pro and a year or more on Enterprise, per the pricing page's feature table.

What is n8n used for?

n8n is used for the glue work between systems. Typical jobs:

• moving a lead from a form into a CRM and a Slack channel;

• turning an email attachment into a database row;

• posting a daily report from a spreadsheet or an analytics API;

• marketing and content automation: drafting social posts from a feed, publishing to a CMS, syncing a mailing list;

• enriching a signup with a third-party lookup;

• running an AI step over incoming support tickets.

The template library's own categories are AI, Sales, IT Ops, Marketing and Document Ops, which is a fair map of where people use it.

The ranking guides for this query describe use cases in the abstract. One concrete case: the editors of this site run an n8n instance behind an internal SEO tool. Seven workflows each sit behind a webhook. They cover keyword discovery against the Ahrefs API, a SERP gap check, a competitor gap check, a drafting step that calls an OpenAI model, internal linking through a WordPress site's API, schema markup and an export. The tool itself is one HTML page that calls those webhooks. That is a typical n8n deployment: a thin front end, the logic in workflows, every outside service reached through an HTTP Request or app node.

Three limits are worth knowing before you plan around it. Heavy or long-running jobs belong in queue mode with workers, not in a single process, which n8n's hosting docs cover. The free Community edition has no projects, no sharing of workflows or credentials between users and no Git version control; those are paid features, listed below. And the AI nodes still need a model provider. n8n Cloud includes "AI models without API keys" on its plans; a self-hosted instance uses your own OpenAI, Anthropic, Google or local-model key.

Is n8n free to use?

Yes, if you host it yourself. The self-hosted Community edition is free to download and run, with no time limit and no execution cap, under the Sustainable Use License described below. There is no free plan on n8n Cloud. The cloud trial docs give you "14 days to try all the features of the Pro plan", capped at 1,000 executions. After that you pick a paid plan or the instance stops.

The cloud trial page of n8n's docs, 30 October 2026: 14 days of Pro features, 1,000 executions, then a paid plan.

So can you use n8n for free for life? On your own hardware, yes. n8n's community edition page says you can "use the Community edition indefinitely", and that it "includes almost the complete feature set of n8n". What it leaves out:

• custom variables, environments and external secrets;

• external storage for binary data and log streaming;

• multi-main mode (queue mode is included);

• projects, and sharing of workflows and credentials between users;

• SSO through SAML or LDAP;

• version control using Git.

Registering your email unlocks a few extras and stays free.

n8n's docs on 30 October 2026: what the free Community edition leaves out.

The pricing page says the same in one line under its plans: "A standard, self-hosted version of n8n is available on GitHub", next to the repository's star count.

The pricing page's own line on the free, self-hosted Community Edition, 30 October 2026.

"Free" still has a cost: a machine to run it on and someone to update it. A laptop works for learning. For anything that must run while the laptop is shut, people use a small virtual server. That is why hosting companies such as OVHcloud and MilesWeb publish their own "what is n8n" pages, two of which Google's AI Overview cites for this query.

How much does n8n cost a month?

On 30 October 2026, n8n's pricing page, loaded from Delhi, showed three cloud plans priced in euros. Starter was €20 a month and Pro €50 a month, both "billed annually". Business was €667 a month for self-hosted use, and Enterprise on request. The annual toggle says "Save 17%", so paying monthly costs more. The page showed no rupee price.

n8n's pricing page on 30 October 2026, loaded from Delhi: euros, no rupee price, billed per execution.

The unit matters more than the price. n8n bills per execution, and its pricing page defines one as "a single run of your entire workflow. It doesn't matter how many steps are in the workflow or how much data it processes." Zapier and Make count every step.

On Zapier's pricing page the same day, the free plan allowed 100 tasks a month and Professional started at $19.99 a month. A task is one action in one Zap, so a Zap with four actions that runs 500 times uses 2,000 tasks. Make's free plan gives 1,000 credits a month, one credit per operation. On n8n Starter, 500 runs of a five-step workflow use 500 of your 2,500 executions.

Zapier's pricing page on 30 October 2026, for the per-task comparison.

Two more lines from the pricing page. Startups with under 20 employees can apply for 50% off Business. And "If it doesn't run, you don't pay": you are billed only "when a workflow runs from start to finish".

What is the n8n AI agent used for?

The n8n AI agent is used to let a language model decide which steps to take, instead of you fixing the order. In the AI Agent node you connect a chat model (OpenAI, Anthropic, Google Gemini or a local model), one or more tools, and optionally memory. Tools can be an HTTP request, a Google Sheets lookup, a calculator or another workflow. The AI Agent node docs call it a system that "receives data, makes rational decisions, and acts within its environment", and say you "must connect at least one tool sub-node".

The AI Agent node page of n8n's docs, 30 October 2026.

In practice it is the node for messy input, where the right action depends on reading it. A support email might need a refund, a reply or an escalation. A document might be an invoice or a contract. A chat question might have its answer spread across three systems. A fixed workflow handles the known cases; the agent handles the judgement calls, using the tools you gave it. n8n contrasts the agent with an "AI chain", which calls a model in a fixed sequence and does not choose its own steps.

Three facts from the docs that the ranking guides skip:

• The node's older "agent type" setting was deprecated in n8n 1.82.0, when every AI Agent became a Tools Agent. The docs say the v1 node "will be removed from n8n 3.0", so an old workflow needs updating before an upgrade.

• The MCP Server Trigger node, listed on every Cloud plan, exposes an n8n workflow as a tool that an outside AI client can call over the Model Context Protocol.

• Cloud plans include monthly Assistant credits (1,600 on Starter, up to 9,600 on Pro) for n8n's own in-editor helper, per the pricing page. They are separate from the models your agents call.

n8n Cloud or self-hosted: which should you pick?

n8n Cloud is the pick if you want to build today and never think about servers: n8n hosts your instance, updates it and includes model access, and you pay by execution. Self-hosting is the pick if your data must stay on your own infrastructure, your volume is high, or your budget is zero. The docs' decision page frames it as two questions: who manages the infrastructure, and which plan or edition has the features you need.

How do you install n8n, and what does n8n 3.0 change?

Until this week, the quickest way to try n8n was one terminal command, npx n8n, then opening http://localhost:5678. n8n's npm install page still documents it and requires "a Node.js version between 20.19 and 24.x, inclusive". The page now carries a notice: "npm-based installs are deprecated from n8n 3.0."

The npm install page of n8n's docs on 30 October 2026: the deprecation notice, the Node.js range and the npx command.

n8n 3.0.0 was released on 27 October 2026 as an early access release, three days before we checked. The release notes call it "a major release with breaking changes". The breaking changes page is explicit: "Self-hosted n8n requires a Docker-based deployment." It adds that 3.0 "no longer supports installations run using npm or npx n8n" and "doesn't publish a runnable n8n package to npm".

The replacement is a one-line setup script, curl -fsSL https://get.n8n.io | sh, which checks for Docker, writes the configuration and starts an instance. A team instance uses your own Docker Compose file with Postgres instead of the bundled SQLite.

n8n's release notes on 30 October 2026: the 3.0 entry, dated 27 October.

The stable line is still 2.x. The newest GitHub release on 30 October 2026 was n8n 2.45.7, published that morning. If you run 2.x from npm today, nothing breaks until you upgrade. The breaking-changes page says to move to Docker first, then run the migration tool under Settings, Migration Report, which lists the workflows and settings on your instance that need attention.

We did not run n8n for this guide. Our test machine's Node.js is version 26, outside the documented range for the npm route, and the Docker route needs a daemon that was not running. We say so because the install facts above are n8n's documentation, not a hands-on result.

Is n8n open source?

Not by the usual definition. n8n's code is public on GitHub and free to use, but it is licensed under n8n's own Sustainable Use License, which the company calls "fair-code". The license grants a free right "to use, copy, distribute, make available, and prepare derivative works", with three limits:

• you may use or modify it "only for your own internal business purposes or for non-commercial or personal use";

• you may give it to others "only if you do so free of charge for non-commercial purposes";

• you may not remove or obscure its notices.

Files with ".ee." in the name are under a separate Enterprise License and need a paid key.

The license FAQ in n8n's docs, 30 October 2026: the three limits of the Sustainable Use License.

Those limits are why the Open Source Initiative's definition does not cover it, and why Wikipedia and many developers call n8n "source-available" rather than open source. The project started in 2019 under Apache 2.0 with the Commons Clause. It moved to its own license in March 2022, per n8n's license FAQ, which gives the company's reasoning: free for users, sustainable as a business.

For most readers the practical line is simple. Automating your own company's work, consulting for clients on their instances, and learning on it are all fine. Selling n8n inside a product, or hosting it for other people as a paid service, needs a commercial agreement with n8n.

Who owns n8n, and how big is it?

n8n is owned by n8n GmbH of Berlin, founded by Jan Oberhauser. It had raised about $240 million by October 2025, before SAP's investment. The sources below are the ones Wikipedia's article cites; we read each.

Wikipedia's funding section for n8n on 30 October 2026; the figures were checked against the Bloomberg, TechCrunch and heise reports it cites.

The growth behind those numbers shows on GitHub. Wikipedia records that n8n topped the JavaScript Rising Stars ranking for 2025, gaining over 112,000 stars in one year, the biggest annual gain in that ranking's decade. Our own reading of the repository on 30 October 2026 put it at 209,117 stars.

The n8n repository on GitHub on 30 October 2026. The GitHub API counted 209,117 stars and 61,317 forks that day.

n8n vs Zapier vs Make

n8n is the one of the three you can run yourself, and the one that bills by workflow run. Zapier is the easiest and the most expensive at volume. Make sits between, with a visual builder closer to n8n's and per-operation credits. Our figures are from each company's pricing page on 30 October 2026, linked above.

n8n in India

India is n8n's second-largest search market for this question. Ahrefs' data for "what is n8n" puts India at 7,400 searches a month against 9,600 in the United States, out of 30,000 worldwide. What an Indian reader should know is mostly about money and hosting. n8n Cloud is priced in euros, and the pricing page showed no rupee price from Delhi. A Starter plan costs whatever €20 is on the day plus your card's foreign-transaction charge. The free Community edition on a small Indian virtual server is the common alternative, which is why hosting companies' guides rank for the query.

There is no official n8n mobile app. A search of the Indian App Store through Apple's search API on 30 October 2026 returned ten apps for "n8n". All were from third-party developers, with names such as "n8n Viewer", "Nodey: n8n Workflows" and "N8N Manager", and none from n8n GmbH. n8n is a web application; on a phone you open your instance in the browser.

How this was checked

We read these on 30 October 2026 from Delhi: n8n's pricing page, the GitHub repository and its latest release, the LICENSE file and the license FAQ, and the docs pages on the cloud trial, the community edition, npm installs, the AI Agent node, hosting, privacy and the glossary. We also read the release notes, the 3.0 breaking-changes page, the integrations directory and the template library.

Funding figures come from the TechCrunch, Bloomberg and heise reports that Wikipedia cites. Zapier's and Make's pricing pages were read the same day; Make's blocked our fetcher before its prices rendered. The App Store search used Apple's public search API for the India storefront. We did not install or run n8n, for the reasons given in the install section, and no claim above rests on a hands-on test. Screenshots are crops of the pages named, taken the same day.

Frequently asked questions

What is n8n in simple terms?

n8n is a tool that connects your apps so that tasks run automatically. You draw the steps as boxes on a canvas, such as "when a form is submitted, add a row to a sheet and post to Slack", and n8n runs them for you. It is free to run on your own computer and sold as a hosted service from €20 a month.

What does n8n stand for?

n8n stands for "nodemation", a blend of "node" and "automation", written as an n, eight letters and an n. Its founder, Jan Oberhauser, chose the name when he started the project in 2019.

Is n8n free to use?

Yes, self-hosted: the Community edition is free with no time limit and no execution cap. n8n Cloud, the hosted version, has a 14-day trial and then costs from €20 a month on annual billing.

Can I use n8n for free for life?

On your own machine or server, yes. n8n's docs say you can use the Community edition indefinitely. The paid parts are team features such as projects, sharing, SSO and Git version control, and the hosted Cloud service.

How much does n8n cost a month?

On 30 October 2026 the pricing page showed Starter at €20 a month and Pro at €50 a month, both billed annually. Business was €667 a month for self-hosted teams, and Enterprise on request. Monthly billing costs more; the page says annual billing saves 17 percent.

What is the n8n AI agent used for?

The AI Agent node lets a language model choose which tools to call to finish a task, such as reading a support email and deciding whether to refund, reply or escalate. You connect a chat model, at least one tool and optional memory, and the agent decides the steps at run time.

Is n8n open source?

Its source code is public and free to use, but under the Sustainable Use License, which limits use to internal business, personal and non-commercial purposes. That is not an open-source license by the Open Source Initiative's definition, so the accurate term is source-available or, in n8n's words, fair-code.

Is n8n better than Zapier?

For technical users and high volumes, usually: n8n can be self-hosted for free, allows code in any step and bills per workflow run rather than per task. For non-technical users who want the most app integrations with no setup, Zapier is easier. The honest answer depends on who maintains it.

What changed in n8n 3.0?

n8n 3.0.0, in early access since 27 October 2026, drops npm and npx installs: self-hosted n8n now needs Docker, and n8n no longer publishes a runnable package to npm. Its docs say the old v1 AI Agent node goes with it, and it adds a migration report under Settings to list what needs attention before you upgrade.