n8n is grade 3 of 3 on all four capabilities this page grades: triggers and app actions, its AI Agent node, its MCP Server Trigger, and self-hosting with a public API. It is free as the self-hosted Community edition. n8n Cloud starts at 20€ a month on Starter, billed annually, for 2,500 workflow executions with unlimited steps. The verdict: a strong pick for teams that want automations and AI agents they can run on their own servers.
What changed: Nov 3, 2026: first published; n8n graded on four capabilities from its pricing page and docs.
n8n review: grade 3 for integrations and AI agents
n8n is grade 3 for integrations and AI agents, from 20€/mo on Starter billed annually; the self-hosted Community edition is free (n8n pricing and n8n hosting docs, checked Nov 3, 2026).
n8n lists "2,000+ integrations for apps, services, and APIs". Its AI Agent node picks which tools to call on its own. Its MCP Server Trigger makes n8n tools and workflows available to outside MCP clients. And you can run all of it on your own infrastructure.
How we grade
Each capability gets a depth grade from 0 to 3. Capabilities are shelved by grade, then plan required, then price. A tie that survives all three steps is listed in alphabetical order, which says nothing about which is better. Plan required runs lowest first: free, then the entry paid plan, then higher plans, then Enterprise or sales only. Price is the lowest list price for that plan, with its billing period, as n8n shows it. "Not documented" never counts as "no".
| Grade | Triggers and app actions | AI agent and tools | MCP | Self-hosting and API |
|---|---|---|---|---|
| 0 | None | None | None | None |
| 1 | Single step only | AI only through an HTTP call to a model provider | MCP only through a generic HTTP call | One of: inbound webhooks, public API, self-hosting |
| 2 | Multi-step workflows, no webhook triggers | AI steps inside workflows, no agent | MCP client tools for agents only | Two of the three |
| 3 | Multi-step workflows plus webhook triggers over a native catalog | An agent node that picks and calls tools on its own | A built-in MCP server that exposes the platform's own workflows to outside MCP clients | All three |
n8n at a glance
Shelved by depth grade, then plan required, then price. Last checked: Nov 3, 2026.
| Capability | Grade | Plan required and price | What n8n's pages say | Last checked |
|---|---|---|---|---|
| Self-hosting and API | 3 | Free: Community edition, self-hosted | Self-host "on your own infrastructure, on-premises, or in a private cloud"; public API on all self-hosted editions; Webhook node | Nov 3, 2026 |
| AI agent and tools | 3 | Starter, 20€/mo billed annually (node docs name no plan; pricing says every integration on all paid plans) | "the agent decides which tools to call to complete a task" | Nov 3, 2026 |
| MCP | 3 | Starter, 20€/mo billed annually (node docs name no plan; pricing says every integration on all paid plans) | MCP Server Trigger makes "n8n tools and workflows available to MCP clients" | Nov 3, 2026 |
| Triggers and app actions | 3 | Starter, 20€/mo billed annually | "2,000+ integrations"; Webhook node; executions "with unlimited steps" | Nov 3, 2026 |
All four tie at grade 3. The Community edition is free, so self-hosting and API comes first; the other three tie on Starter at 20€/mo billed annually and are listed alphabetically.
Self-hosting, API and security (grade 3)
n8n is grade 3 for self-hosting and API, free as the self-hosted Community edition (n8n hosting docs, checked Nov 3, 2026).
You can self-host n8n on your own infrastructure, on-premises or in a private cloud. Without a license key, n8n runs as the free Community edition. The one-line setup requires Docker on Linux or macOS, or WSL on Windows. Docker Compose is the route for production deployments with databases and other services. npm installation is deprecated from n8n 3.0.
The public API lets you do many of the same tasks you do in the editor, from code. It is on all self-hosted editions and on the Starter, Pro and Enterprise cloud plans. It is not on the free trial.
n8n's security docs for a self-hosted instance cover:
- SSL to enforce secure connections.
- SAML or OIDC single sign-on, on Business or Enterprise, not the Community edition.
- 2FA for your users.
- Turning off the public REST API.
- Opting out of the anonymous data n8n collects automatically.
- Blocking specific nodes so users cannot reach them.
You set these up yourself on a self-hosted instance.
- Best for: teams that want automations and agents on their own servers, with a public API on every self-hosted edition.
- Limitation: someone has to run and secure the instance, and the one-line setup needs Docker.
- Plan required: Free (Community edition, self-hosted). Price: free.
- Last checked: Nov 3, 2026.
AI Agent node and tools (grade 3)
n8n is grade 3 for AI agents with tools, from 20€/mo on Starter billed annually (n8n AI Agent node docs, checked Nov 3, 2026).
The AI Agent node lets you build an AI agent inside a workflow. You connect a chat model and one or more tools. The agent then decides which tools to call to finish the task. You must connect at least one tool sub-node.
- Best for: agents that choose their own next step across the apps in a workflow.
- Limitation: the agent needs at least one tool sub-node, and the node docs name no plan.
- Plan required: Starter (every integration on all paid plans). Price: 20€/mo, billed annually.
- Last checked: Nov 3, 2026.
MCP Server Trigger (grade 3)
n8n is grade 3 for MCP, from 20€/mo on Starter billed annually (n8n MCP Server Trigger docs, checked Nov 3, 2026).
The MCP Server Trigger lets n8n act as a Model Context Protocol server. Outside MCP clients can list the available tools and call them one at a time. The trigger only connects to and runs tool nodes. To expose a whole workflow, you wrap it in the Custom n8n Workflow Tool node.
Transport matters. The trigger supports Server-Sent Events (SSE) and streamable HTTP. It does not support standard input/output (stdio). In queue mode with several webhook replicas, route every /mcp* request to one dedicated replica, or SSE and streamable HTTP connections may break. Behind a reverse proxy such as nginx, turn off proxy buffering for the MCP endpoint.
- Best for: exposing your own n8n workflows as tools to an MCP client.
- Limitation: no stdio transport, and queue mode needs all MCP traffic on a single replica.
- Plan required: Starter (every integration on all paid plans). Price: 20€/mo, billed annually.
- Last checked: Nov 3, 2026.
Triggers and app actions (grade 3)
n8n is grade 3 for triggers and app actions, from 20€/mo on Starter billed annually (n8n integrations, checked Nov 3, 2026).
The catalog lists 2,000+ integrations for apps, services and APIs, plus an HTTP Request entry. The Webhook node creates webhooks that "can receive data from apps and services when an event occurs." It accepts DELETE, GET, HEAD, PATCH, POST and PUT requests, and the maximum payload is 16MB. n8n registers the production webhook when you publish the workflow.
n8n's plans count executions, not steps. The pricing page says "An execution is a single run of your entire workflow," however many steps it has or how much data it processes.
- Best for: event-driven workflows with many steps, since a long workflow still counts as one execution.
- Limitation: a webhook payload tops out at 16MB.
- Plan required: Starter. Price: 20€/mo, billed annually.
- Last checked: Nov 3, 2026.
Plans and executions
Starter, Pro, Business and Enterprise all include unlimited users, unlimited workflows and every integration. Plans differ by executions, concurrency and hosting.
Plans shown as n8n lists them, in EUR, billed annually. Last checked: Nov 3, 2026.
| Plan | Price | Workflow executions | Concurrent executions | Hosting | Last checked |
|---|---|---|---|---|---|
| Community edition | Free | Not documented | Not documented | Self-hosted | Nov 3, 2026 |
| Starter | 20€/mo, billed annually | 2.5K, unlimited steps | 5 | Hosted by n8n | Nov 3, 2026 |
| Pro | 50€/mo, billed annually | 10K, unlimited steps | Up to 50 | Hosted by n8n | Nov 3, 2026 |
| Business | 667€/mo, billed annually | 40K, unlimited steps | Not documented | Self-hosted only | Nov 3, 2026 |
| Enterprise | Contact Sales | Custom | 200+ | Hosted by n8n or self-hosted | Nov 3, 2026 |
Starter includes 1,600 Assistant credits a month, and Pro up to 9,600. Business adds SSO and LDAP, multiple environments, queue mode for scaling and version control using Git.
The trial needs no credit card. Its limits are 1,000 executions, 5 concurrent executions, a 180-second execution timeout and 1 team project. A Business trial gives full access to Business features for 14 days.
Who should skip it
- Teams that want Business on n8n Cloud. n8n says the Business plan is currently only available for self-hosted n8n.
- Teams that need the public API during the trial. The API is not available on the free trial.
- Teams whose MCP client only speaks stdio. The MCP Server Trigger supports SSE and streamable HTTP only.
- Teams with no one to run Docker. Self-hosting is the free route, and the one-line setup requires Docker.
Best for
- Best for self-hosting: n8n Community edition (grade 3, free, self-hosted).
- Best for AI agents with tools: n8n AI Agent node (grade 3, Starter, 20€/mo billed annually).
- Best for exposing workflows over MCP: n8n MCP Server Trigger (grade 3, Starter, 20€/mo billed annually).
- Best for SSO and Git version control: n8n Business (667€/mo billed annually, self-hosted only).
FAQ
Is n8n free?
Yes, if you self-host: without a license key, n8n runs as the free Community edition. n8n Cloud also has a trial with 1,000 executions and no credit card.
How much does n8n cost?
n8n Cloud costs 20€ a month on Starter and 50€ on Pro, billed annually. Business is 667€ a month billed annually and self-hosted only, and Enterprise is priced by sales.
Is n8n good for beginners?
n8n publishes no ease-of-use rating. A beginner can start on the cloud trial with no credit card, while self-hosting asks for Docker.
Is n8n safe to self-host?
n8n documents SSL, 2FA, turning off the public API and blocking nodes for self-hosted instances, plus SAML or OIDC single sign-on on Business or Enterprise. The operator sets each of these up.
What counts as an execution in n8n?
An execution is a single run of your entire workflow. The number of steps and the amount of data it processes do not change the count.
Does n8n have AI agents?
Yes. The AI Agent node connects a chat model and one or more tools, and the agent decides which tools to call to complete a task.
Does n8n support MCP?
Yes. The MCP Server Trigger makes n8n tools and workflows available to MCP clients over SSE or streamable HTTP, but not stdio.
How many integrations does n8n have?
n8n lists 2,000+ integrations for apps, services and APIs. Every paid plan includes every integration.
Sources
- n8n pricing (checked Nov 3, 2026)
- n8n integrations (checked Nov 3, 2026)
- n8n Webhook node docs (checked Nov 3, 2026)
- n8n AI Agent node docs (checked Nov 3, 2026)
- n8n MCP Server Trigger docs (checked Nov 3, 2026)
- n8n hosting docs (checked Nov 3, 2026)
- n8n public API docs (checked Nov 3, 2026)
- Securing n8n (checked Nov 3, 2026)
- n8n single sign-on docs (checked Nov 3, 2026)
- n8n Community edition features (checked Nov 3, 2026)