n8n (pronounced “n-eight-n”) is a workflow automation tool with an unusual position in the market. It is a visual, node-based builder like its better known rivals, but its documentation describes it as fair-code licensed, a self-hosted version is available, and its pricing is based on workflow executions rather than on individual steps. In recent years it has added a set of AI features, including an agent node and MCP support.
This guide profiles n8n using the vendor’s own pricing page, security page and documentation: what it is, who it suits, the features it documents, how its plans work, where it is strong and where it is limited, and how it compares with Zapier and Make. For a quicker read, see our shorter review.
How we researched this: we did not run hands-on tests for this article. It is based on the vendor’s own documentation and public sources, checked on 14 November 2026. Features and prices change, so confirm details on the vendor’s site before you buy.
Key takeaways
- n8n’s documentation describes it as a fair-code licensed workflow automation tool that combines AI features with business process automation.
- Pricing is based on monthly workflow executions, regardless of complexity. On 14 November 2026 the pricing page showed Starter at 20 euros per month, Pro at 50 euros and Business at 667 euros, all billed annually.
- All plans include unlimited users and workflows and every integration, according to the pricing page, and a Community Edition can be self-hosted.
- The AI toolkit includes an AI Agent node, an MCP Server Trigger, an MCP Client, human approval for tool calls and hosted chat.
- Self-hosting gives control but also responsibility: n8n’s security page says self-hosters must handle TLS and encryption at rest themselves.
What n8n is and who makes it
n8n is a workflow automation tool made by n8n, a company that describes itself on its security page as based in Berlin. The docs home page calls n8n “a fair-code licensed workflow automation tool that combines AI features with business process automation.” You build workflows in an editor by connecting nodes, run them from triggers such as webhooks or schedules, and can add code steps when visual nodes are not enough.
The company offers n8n in several forms. There is a cloud service hosted by n8n, a Community Edition that you can run yourself from the public repository, and paid self-hosted plans for businesses and enterprises. The documentation offers one-line setup with Docker Compose, deployment on a cloud provider, or a free trial followed by a plan. Our attempts to load the main n8n.io homepage in the browser we used timed out, so the screenshot in this article shows the pricing page instead.
Who n8n is for
- Technical teams and developers. The pricing page lists code steps in JavaScript or Python, custom API requests including HTTP and GraphQL, cURL import, control through an API and a command line, and (on self-hosted) bash scripts and custom nodes.
- Organisations that want to self-host. A Community Edition and self-hosted Business and Enterprise plans exist, and the security page explains what self-hosters must handle.
- Teams building AI agents. The AI Agent node, MCP nodes and evaluation tools are listed across plans.
- Cost-conscious teams with complex workflows. The pricing page says n8n charges per workflow run rather than per step or user, which can matter if your workflows have many steps.
Non-technical users may find it a steeper climb than simpler tools. Our article n8n vs Zapier: self-hosted or hosted explores that trade-off.
Core features
Visual editor and workflow templates
The pricing page lists a “leading editor UI” and workflow templates on every plan, together with all integrations, which it calls nodes. It does not give a count of nodes, so we could not confirm how many integrations exist today.
Code steps and custom requests
Every plan includes code steps in JavaScript or Python, custom API requests (HTTP, GraphQL and others), cURL import, webhook and queue triggers, and the ability to control n8n through an API and CLI. This is the main reason developers consider n8n.
Data tables
Data tables are listed as included on all cloud plans, giving workflows a place to store structured data.
Debugging and error handling
The page lists execution logging, error workflows, debugging in the editor and automatic retries on all plans. Starter does not include searching executions by the data they contain, which appears from Pro upward.
Workflow history
The comparison table lists workflow history for a maximum of 1 day on Starter, 5 days on Pro and 365 or more days on Enterprise. If you rely on being able to roll back changes, check this row closely.
Credentials and collaboration
The security section of the pricing table lists an encrypted credential store and credential sharing on all plans, plus AI usage controls. Collaboration is organised by projects: Starter has one shared project with project admins, Pro adds project editors, instance admins and three shared projects, and Enterprise adds project viewers and custom roles. Users are unlimited on every plan, so the practical limits are projects and roles rather than seats.
Insights
An insights overview banner is included on all plans. Pro adds a table with workflow metrics and a dashboard showing change over time, with date filters of seven to fourteen days, while Enterprise lists filters from 24 hours to 365 days and hourly, daily or weekly granularity. This is useful if you need to show how often workflows run and fail.
Newer and advanced features and integrations
AI Agent node and AI steps
The pricing page lists agents (marked as preview), an AI Agent node and AI steps as included in Starter, Pro and Enterprise. Starter lists 1,600 Assistant credits per month, Pro up to 9,600, and “AI models without API keys” appears under Starter.
MCP support
The page lists an MCP Server Trigger, an MCP Client and instance-level MCP on all plans. The docs home page also gives commands for connecting Claude Desktop, Claude Code and the OpenAI Codex CLI to an n8n MCP server. Read the related documentation before exposing workflows to an assistant.
Human approval for tool calls and chat
Human approval for tool calls and hosted and embedded chat are both listed. Approval steps are a practical control when an agent could take an action with real consequences.
Evaluations and tracing
Light evaluations are included on all plans, while metric-based evaluations are limited to one workflow on Starter and unlimited on Pro and Enterprise. LangSmith tracing and OpenTelemetry tracing (preview) are marked as available in self-hosted setups.
Enterprise scaling and governance
Enterprise lists queue mode, worker view, environments, Git version control, workflow diff, audit logging, log streaming to services such as Datadog, external secret store integration, custom roles and enforced two-factor authentication. The Business card also lists queue mode, multiple environments and Git version control on self-hosted deployments.
Pricing and plans
Checked on 14 November 2026 on the n8n pricing page. Prices are shown in euros, and the figures below are the annual-billing prices displayed with the “Annually (Save 17%)” toggle. Prices change, so confirm on the vendor page before you buy.
| Plan | Price as shown | Workflow executions | Where it runs |
|---|---|---|---|
| Starter | 20 euros per month, billed annually | 2,500 per month | Hosted by n8n |
| Pro | 50 euros per month, billed annually | 10,000 per month | Hosted by n8n |
| Business | 667 euros per month, billed annually | 40,000 per month | Self-hosted (the card says self-hosted only) |
| Enterprise | Contact sales | Custom | Hosted by n8n or self-hosted |
The page says all plans include unlimited users and workflows and every integration, with pricing based on monthly workflow executions regardless of complexity. Starter and Pro offer a free trial with no credit card required. The Business card is aimed at companies with fewer than 100 employees, and a Startup plan offers 50 percent off Business to companies with fewer than 20 employees if they qualify. The Community Edition is described as a standard self-hosted version available on GitHub.
Limits that differ by plan
| Limit | Starter | Pro | Enterprise |
|---|---|---|---|
| Max saved executions | 2.5K | 25K | 50K |
| Max execution log retention | 7 days | 30 days | Unlimited |
| Concurrent executions | 5 | 20 or 50 | 200+ |
| Max execution duration | 5 minutes | 40 minutes | 40 minutes |
| Shared projects | 1 | 3 | Unlimited |
| Overage fees | Capped | Capped | Custom |
These limits matter more than they look. A five-minute maximum duration on Starter and five concurrent runs could constrain long or bursty workflows. To turn executions into a forecast, follow the method in our guide to reading an AI tool’s pricing page. We could not confirm from the page how the Business plan’s executions are metered for self-hosted installations, since the page has a separate question about this in its FAQ that we did not read in full.
Strengths
Pricing that does not penalise long workflows
The page states that n8n charges when a workflow runs from start to finish, not per step or per user. For a workflow with many steps, that is a different cost shape from per-action billing.
Self-hosting option
A Community Edition and self-hosted plans give you control over where data and workflows live. Our data protection guide explains why that can matter.
Depth for developers
Code steps, custom requests, API and CLI control, and (when self-hosted) custom nodes and bash scripts give technical teams a lot of room.
Rich AI toolkit on every cloud plan
The Agents row, MCP support, human approval and chat appear on Starter, Pro and Enterprise alike, so you do not need the top tier to experiment with agents.
Transparent plan comparison
The pricing page lists a long feature-by-plan table, so you can see exactly what is missing from each tier, including negatives such as “Not included”.
Weaknesses and limits
Technical learning curve
The very features that attract developers, such as expressions, code steps and flow logic, mean that non-technical colleagues may need training. The docs list expressions, flow logic and executions as key concepts to learn.
Starter has tight limits
Five concurrent executions, a five-minute maximum duration, seven days of log retention and one shared project are modest if you plan to run many automations.
Version control and environments are Enterprise features
The comparison table shows Git version control, environments and workflow diff as not included on Starter and Pro. Teams that need a proper development and production process will need a higher tier.
SSO and audit logging are gated
SSO and LDAP, audit logging, log streaming, enforced two-factor authentication and external secret stores are shown as not included on Starter and Pro. The Business card lists SSO and LDAP for self-hosted use.
A big step to Business
The gap between Pro at 50 euros and Business at 667 euros per month is large, and Business is self-hosted only on the card we read. If you want hosted collaboration features at a middle price, that may not exist.
Self-hosting means you run it
The security page says self-hosters must set up a reverse proxy for TLS and handle encryption at rest. That is more work than a hosted service.
Privacy, data handling and security
From n8n’s security page, read on 14 November 2026:
- n8n says it aligns its security programme to SOC 2, with continuous evaluation and annual audits by an independent auditor. The SOC 2 report is available to enterprise customers, and others can use the SOC 3 report.
- n8n Cloud uses Microsoft Azure, and the page says the hardware and data are currently hosted in the European Union, with plans for more locations.
- Data at rest on cloud is encrypted using Azure server-side encryption (AES256, FIPS-140-2 compliant implementation), and traffic is encrypted in transit.
- Cloud customer instances are logically isolated, credentials are not logged or exported by default, and execution data is deleted according to retention settings.
- Authentication uses a username and password, with MFA optional. SSO, SAML and LDAP are listed for the Enterprise plan on that page, while the pricing page lists SSO and LDAP on Business as well.
- Audit log history is kept for at least 12 months, backups run daily and are replicated to a separate region in the same country, and third-party penetration tests happen at least annually.
- For self-hosted installations, encryption in transit and at rest is the user’s responsibility.
We did not read n8n’s statements about AI model training or its sub-processors list, so we cannot report on them. Note also that our two pages differ slightly on which tiers include SSO, so confirm with n8n.
How n8n compares
| Point | n8n | Zapier | Make |
|---|---|---|---|
| Billing unit | Workflow executions | Tasks | Credits |
| Entry paid price shown | 20 euros per month billed annually | From $19.99 per month | $9 per month |
| Free option | Trial on cloud, self-hosted Community Edition | Free plan, 100 tasks per month | Free plan, 1,000 credits per month |
| Self-hosting | Yes | Not described in pages we read | Not described in pages we read |
| Code steps | JavaScript and Python on all plans | Not covered in our reading | JavaScript or Python mentioned on homepage |
| Stated compliance | Aligned to SOC 2, SOC 3 report available | SOC 2 Type II, SOC 3 | SOC 2 Type II, SOC 3, GDPR |
For comparisons in more depth, see n8n vs Zapier and Zapier vs Make. If your interest is in agents more broadly, our piece AI agents: what is real and what is hype offers context.
Getting started
- Decide whether you want cloud or self-hosted. If unsure, start with the cloud free trial, which needs no credit card according to the pricing page.
- Follow the docs to build a first workflow, such as an API endpoint or a scheduled job.
- Learn the key concepts the docs highlight: expressions, flow logic and executions.
- Count how many executions a month your real process needs, using the execution definition on the pricing page.
- If you test agents, add a human approval step before any action with real consequences.
- Check the limits for the plan you are considering, especially concurrency and duration.
Frequently asked questions
Is n8n free?
There is a free self-hosted Community Edition on GitHub according to the pricing page, and a free trial of the cloud plans with no credit card required. Hosted plans are paid.
What counts as an execution?
The pricing page says you pay when a workflow runs from start to finish, and that pricing is based on monthly workflow executions regardless of complexity.
Does n8n charge per user?
No. The page states that all plans include unlimited users and workflows.
Where is n8n Cloud data stored?
The security page says n8n Cloud uses Microsoft Azure with data currently hosted in the European Union.
Does n8n support AI agents and MCP?
Yes. The pricing page lists an AI Agent node, an MCP Server Trigger, an MCP Client and instance-level MCP, and the docs explain how to connect assistants such as Claude Desktop.
Is n8n open source?
The docs call n8n fair-code licensed. We did not read the licence text, so check its terms before using it in a commercial product.
Is n8n SOC 2 certified?
The security page says n8n aligns its programme to SOC 2, with annual audits by an independent auditor, and offers a SOC 3 report.
How much support do you get?
Every plan includes access to the forum. Email support and dedicated support with an SLA are listed for Enterprise only, and the Business card also lists forum support. If you expect to need a person to answer production incidents, factor that into the tier you choose, and read the pricing page FAQ on the difference between forum and dedicated support.
Conclusion
n8n is a good fit for technical teams that value self-hosting, code-level control and an execution-based pricing model that does not penalise long workflows. Its AI features are broad and available on all cloud plans, but its smaller plans have tight limits and some governance features sit on the highest tiers. Non-technical teams, or those that want a simple hosted experience, may be better served by Zapier or Make. Run a trial against a real workflow, measure executions, and use our tool choice guide to document your decision.
Sources
- n8n pricing, accessed 14 November 2026
- n8n security, accessed 14 November 2026
- n8n docs home, accessed 14 November 2026
- Zapier pricing, accessed 14 November 2026
- Make pricing, accessed 14 November 2026
