# What the n8n AI Agent node is and what it costs

n8n's AI Agent node needs a chat model and at least one tool, and its AI category holds 9,723 community templates. n8n counts a run as one execution however many steps it has, and says it is not open source. Read 2026-11-06.

Short answer, from the sources below: an n8n AI agent is one node that is given a chat model and tools, and the templates for it are community-made workflows that each ask for their own model key. n8n bills by workflow execution, not by step, and says the model's cost is outside that. Its licence allows use for a business's own purposes and for building workflows for others, and does not allow offering n8n to customers to build their own. This page lists what each source says, who wrote it and what it leaves out. The prices of the plans are on the [no-code builders page](/research/no-code-ai-agent-builders).

## What the AI Agent node is

n8n's documentation says: "The AI Agent node lets you build an AI agent in n8n. Connect a chat model and one or more tools, and the agent decides which tools to call to complete a task."[1] It adds: "You must connect at least one tool sub-node to an AI Agent node."[1] Since n8n 1.82.0 the node's agent type setting is deprecated, and "All AI Agent nodes now work as a Tools Agent." The older version of the node (v1) "will be removed from n8n 3.0."[1]

## What the template library holds

n8n's AI category page is titled "Discover 9723 AI Automation Workflows from the n8n's Community", and the page calls them "Top 9,723 AI workflow automation templates".[2] Its featured list mixes chatbots with document and email work: "Chat with a database using AI", "RAG chatbot for company documents using Google Drive and Gemini", "Basic automatic Gmail email labelling with OpenAI and Gmail API" and "Rank supplier RFPs with Groq LLM, Airtable and Slack".[2] Each template names its author; the three read for this page were each marked "Last update 8 months ago".[3][4][5]

Three templates, read in full:

- **Build your first AI agent**, by Lucas Peyrin. It is "an AI-powered chatbot that can do more than just talk", wired to Google Gemini with two tools, "Get Weather" and "Get News" (an RSS reader), and a conversation memory node. Its setup time is "~2 minutes", and the one thing it needs is "a free Google AI API key."[3]
- **Chat with a database using AI**, by David Roberts. It answers questions about data in Postgres, MySQL or SQLite. "To use it, you'll need an OpenAI API key (although you could also swap in a model from another service)", and it needs "n8n version 1.19.4 or later."[4]
- **AI invoice agent**, by Rakin Jakaria. Despite the name, it makes and sends invoices; it does not extract data from them. It "automates the invoice creation, email delivery, and status tracking process for client billing": a manual trigger, a Google Sheet of invoices with Status = Pending, GPT-4.1-mini writing the email, CraftMyPDF building the PDF, Gmail sending it, and the sheet updated to Completed.[5] The model's part in that sequence is writing the email; the steps around it are fixed.[5]

## What running one costs

n8n's pricing page says: "Pricing is based on monthly workflow executions, regardless of complexity."[6] Its FAQ defines an execution: "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."[6] The Starter plan is 20 euros a month billed annually for 2.5k workflow executions, and lists "Agents, MCP and AI nodes".[6] The FAQ does not say how a run in which an agent calls several tools is counted.[6]

The model is billed separately in the templates read: each asks for the provider's key (Google AI, OpenAI), and the pricing page states no price for a model call.[3][4][6] The Starter plan lists "AI models without API keys" and the page does not explain the line.[6] The "Assistant credits" it lists (1,600 a month on Starter) belong to n8n's builder assistant: "Credits are consumed as you use Assistant."[6]

For checking what an agent did, the same page lists "Max execution log retention" as 7 days on Starter, 30 days on Pro and unlimited on Enterprise, and its FAQ says n8n keeps executions "until you reach the first of those, then remove the oldest."[6] The [monitoring and audit trails page](/research/ai-agent-monitoring-audit-trails) sets that beside the retention of other tools.

## What the licence allows

n8n's community-licence page says the source is under the Sustainable Use License: "You may use or modify the software only for your own internal business purposes or for non-commercial or personal use."[7] n8n says of itself: "we do not call ourselves open source", because "open source licenses can't include limitations on use."[7] Building workflows for others is allowed: "you are now free to offer commercial consulting or support services (e.g. building n8n workflows) without the need for a separate license agreement with us."[7] What it does not allow is "making n8n available to your customers for them to connect their accounts and build workflows."[7] The page does not address a case in between, such as running a client's workflows on a server the consultant hosts.[7]

## Not yet proven

- **That any template runs today.** The three read were last updated 8 months before 2026-11-06, and none was run or imported.
- **What a run costs in model fees.** The templates name OpenAI, Google Gemini and GPT-4.1-mini; none of the pages read gives a price per run.
- **How an agent run is counted.** The execution FAQ defines a run as one pass of the whole workflow and does not mention an agent's tool calls.
- **What "AI models without API keys" covers.** The pricing page lists it on Starter without a definition.
- **CraftMyPDF's cost.** The invoice template depends on it and its pricing was not read.
- **Where hosting for a client falls under the licence.** The licence page names two cases, internal use and offering n8n to customers, and says nothing of the case between them.
- **Other template libraries.** Third-party collections of n8n templates were not read.

If you build agents on n8n for clients and have worked out which side of the licence your hosting falls on, I would like to compare notes: mackenzie@mackenziebowes.com.

## Sources

1. n8n Docs, AI Agent node. Read in Rotunda 2026-11-06. <https://docs.n8n.io/integrations/builtin/cluster-nodes/root-nodes/n8n-nodes-langchain.agent>
2. n8n, AI Automation Workflows templates category. Read in Rotunda 2026-11-06. <https://n8n.io/workflows/categories/ai/>
3. n8n template 6270, Build your first AI agent, by Lucas Peyrin. Read in Rotunda 2026-11-06. <https://n8n.io/workflows/6270-build-your-first-ai-agent>
4. n8n template 2090, Chat with a database using AI, by David Roberts. Read in Rotunda 2026-11-06. <https://n8n.io/workflows/2090-chat-with-a-database-using-ai/>
5. n8n template 7905, AI invoice agent, by Rakin Jakaria. Read in Rotunda 2026-11-06. <https://n8n.io/workflows/7905-ai-invoice-agent/>
6. n8n, Plans and Pricing, including the FAQ answers opened by clicking each question. Read in Rotunda 2026-11-06. <https://n8n.io/pricing/>
7. n8n Docs, Community license. Read in Rotunda 2026-11-06. <https://docs.n8n.io/n8n-community-license>

---

Published 2026-11-06. Web page: https://www.mackenziebowes.com/research/n8n-ai-agents
