Activepieces vs n8n for self-hosting in 2026: license, free plan and your own AI keys

Choosing Activepieces vs n8n to self-host comes down to three things their own pages answer: what the license lets you do with the code, what the free self-hosted edition includes, and how each tool takes your own AI provider keys. Activepieces releases its core under the MIT license and keeps its enterprise features in separately licensed folders. n8n uses its Sustainable Use License, which allows use "only for your own internal business purposes or for non-commercial or personal use", and keeps its enterprise files under a separate n8n Enterprise License. Both document a Docker route, and both can run AI steps on a key you own.

Everything on this page about Activepieces and n8n comes from their documentation, license files and pricing pages as we read them on October 16, 2026. We did not install either tool for this guide or take screenshots, and this page offers no legal advice. For a business decision, read the license file itself and ask the vendor or your lawyer. We make Barside, a free desktop app for Windows, Linux and Mac, and a short part near the end says where it fits and where it does not.

Activepieces vs n8n for self-hosting, side by side

The table sums up what each project's own pages say. The sections below quote the wording and link the page behind every line.

Both tools connect web apps in flows that run on a server, and both document how to host them yourself. The biggest difference for most teams is the license, so start there.

What the Activepieces and n8n licenses allow

Activepieces: an MIT core and a commercial enterprise folder

Activepieces' license page says "Activepieces' core is released as open source under the MIT license." The LICENSE file in its GitHub repository names the exceptions: everything in the packages/ee/ and packages/server/api/src/app/ee directories is under the license in packages/ee/LICENSE, and content outside them is available under the "MIT Expat" license. That MIT text grants permission "free of charge, to any person obtaining a copy" of the software, and the license page names one requirement: include the license notice when you distribute it.

The license page says enterprise and cloud edition features fall under a separate commercial license, and that self-hosted use of those features needs an Activepieces license. The open source page puts it this way: the enterprise code "sits in one clearly marked directory and needs a subscription only for production use."

n8n: the Sustainable Use License

n8n's LICENSE.md file on GitHub grants "a non-exclusive, royalty-free, worldwide, non-sublicensable, non-transferable license to use, copy, distribute, make available, and prepare derivative works of the software", subject to four limitations:

• "You may use or modify the software only for your own internal business purposes or for non-commercial or personal use."
• "You may distribute the software or provide it to others only if you do so free of charge for non-commercial purposes."
• "You may not alter, remove, or obscure any licensing, copyright, or other notices of the licensor in the software."
• "Any use of the licensor's trademarks is subject to applicable law."

The same file says source files with .ee. in their filename or .ee in their folder name are not under the Sustainable Use License, and that using them requires "a valid n8n Enterprise License specifically allowing you access to such source code files." This section quotes the license file in the repository, the text that ships with the code.

What the difference means when you choose

For automations that run your own company's work, both licenses cover that use on their face. The question gets harder when the automation tool itself becomes part of what you sell, such as hosting it for clients or building it into a product. Take that case to the vendor before you build: Activepieces lists Embed and Enterprise options on its pricing page, and n8n lists Business and Enterprise plans on its own. A lawyer who reads the license text with your plan in front of them is the right last step.

Activepieces self-hosted: the official open source edition

The official open source way to run Activepieces self-hosted is its Community edition, which the open source page describes as an "MIT-licensed core, self-hosted with Docker or Helm, with nothing metered." The same page says "Community edition has no credit system and no cap on runs," and that it leaves out "projects, SSO, audit logs and Git Sync." Releases are versioned on Docker Hub and GHCR and ship every Monday, so you can pin a version and upgrade when you choose.

For production, the page lists "A stateless app, one or more stateless workers, PostgreSQL, Redis, and optionally an S3-compatible bucket for run logs." That is the stack you keep up to date, back up and monitor yourself.

Activepieces pricing explains what paying adds to a self-hosted install. Its FAQ calls the Community edition "the automation core on its own," and says paying adds "controlling who has access, seeing who changed what, keeping teams and clients separate, shipping changes safely, and running Agents and Chat." If you need agents or chat on your own server, confirm with Activepieces which plan includes them. The same page lists Enterprise control, with custom pricing, for running Activepieces on your own servers or fully air-gapped, with SSO, SCIM, custom RBAC, audit logs and secret managers.

How to install Activepieces on your own server

Activepieces' install options mark one route as recommended: "Docker Compose with PostgreSQL and Redis."

1. On the machine where Activepieces should live, run the quick install command from the page: curl -fsSL https://get.activepieces.com | sh.
2. Follow the Docker self-hosting guide the page links to for the next steps. The install page says that guide also covers activating a license key, if you buy one.
3. If you already run Kubernetes, the page offers a Helm chart instead, "Only if you already run a cluster."

The page also lists provider templates and one-click hosts, including AWS, GCP, Railway, Elestio, Easypanel, PikaPods, RepoCloud and Zeabur. It warns that they are "Maintained outside the core team, so versions can lag behind, and they install the free Community edition."

n8n Community vs Enterprise features on a self-hosted server

n8n's Community vs Enterprise features for self-hosting are listed on its Compare editions page, which says "Every self-hosted installation runs on the same underlying product." The free Community edition leaves out these features:

• Custom variables
• Environments
• External secrets
• External storage for binary data
• Log streaming
• Multi-main mode
• Projects
• SSO (SAML and LDAP)
• Sharing of workflows and credentials
• Version control using Git

On sharing, the page explains that in the Community edition "Only the instance owner and the user who creates them can access workflows and credentials." It says the missing features "are available on self-hosted Business and Enterprise plans," and that some are also available on n8n Cloud.

There is a free middle step. If you register your Community edition with your email, n8n sends a free license key that unlocks Folders, Debug in editor and Custom execution data. You can choose it during the first account setup, or later in Settings → Usage and plan with Unlock.

n8n's pricing page says "A standard, self-hosted version of n8n is available on GitHub," and marks the Business plan "Self-hosted only." As of October 2026, it lists Business at 667€ a month, billed annually, with 40,000 executions, SSO and LDAP, Git version control and 6 shared projects. Its license key "must ping our license server daily to stay active." Enterprise has no public price.

How to install n8n on your own server

n8n's install options list five routes: a one-line setup from the command line, Docker Compose, npm, Docker, and guides for "DigitalOcean, AWS, Azure, Google Cloud, and more." The page marks none of them as recommended. If you want to run n8n on a Windows computer without containers, our guide to AI automation tools that run on Windows without Docker walks through n8n's npm route and the cautions its docs give.

The Activepieces free tier in 2026, and n8n's cloud trial

The Activepieces free tier in 2026 is the Cloud Free plan. As of October 2026, Activepieces pricing lists it at US$0, "Free forever, no card," with 1,000 credits a month and up to 5 users. Its FAQ explains the meter: "A run is one credit however many steps it has," and "AI is the exception that sits on top." An AI step on a fast model is 2 credits, on a smart model 10, and on a frontier model 20, and each agentic action is 1 credit.

On Free, runs past your quota wait, and the page gives two answers for how long. One FAQ answer says they "do not go through until credits refresh the next day," and the comparison table lists the Free credit reset as "Daily" and going over as "Waits for tomorrow." Another FAQ answer says "You get a fresh 1,000 credits every month, so if you use them all, you wait until the next month." Check the current page before you plan around it.

The paid cloud plans, as of October 2026 with monthly billing (the page says yearly billing saves "2 months"):

• Plus: US$20 "flat per month," with 10,000 credits a month, then pay as you go at US$0.007 per credit, up to 5 users, and "Bring your own AI keys."
• Team: US$200 "flat per month," with 50,000 credits a month, 25 users, SSO and standard roles.
• Ultimate: custom, with "Credits sized to you, yearly."

Embedding Activepieces in your own product is a separate offer, "From $36k a year" as of October 2026.

n8n Cloud works differently. n8n's pricing page lists no free plan as of October 2026, and shows Starter at 20€ a month and Pro at 50€ a month, both billed annually, each with a "Start free trial" button and "No credit card required." The page offers monthly billing too, but the prices written on it are the annual ones. n8n's free trial page says "you have 14 days to try all the features of the Pro plan," with "a limit of 1000 executions." When it ends, "the trial automatically expires and n8n deletes your workspace," and you can download your workflows within 90 days.

An example: counting a month of runs on Activepieces Cloud

This is an example with made-up numbers, to show how the meter adds up. Say a small agency runs a 4-step flow 300 times a month.

• With no AI step, the flow uses 300 credits, since each run is one credit however many steps it has. That fits inside the Free plan's 1,000.
• Add one AI step on a smart model, and the AI steps alone come to 3,000 credits (300 runs at 10 credits), past the Free plan's 1,000 credits a month.
• On Plus with your own AI keys, each AI step drops to 1 credit, so the AI steps come to 300 credits, and your AI provider bills the model use.

How to bring your own API key to Activepieces and n8n

Activepieces

To bring your own API key to Activepieces, a platform admin connects it on the AI Providers page, which carries the subtitle "Bring your own AI keys and use them across every project."

1. Open Platform Admin → Catalogue → AI Center → Providers.
2. Find the provider and choose Connect.
3. Follow the provider's instructions on screen, enter the credentials, and choose Save.

The page lists OpenAI, Anthropic, Google Gemini, Mistral AI, DeepSeek, xAI, Qwen, Z.ai, MiniMax and Moonshot AI, the cloud platforms Azure and AWS Bedrock, the gateways OpenRouter and Cloudflare AI Gateway, and "Other (OpenAI Compatible)" for anything else. You can add several providers and "pick the model per agent or per step." It suggests a gateway in front of your key "when you want limits and an audit trail," and names LiteLLM, Portkey and Helicone as gateways that work through the OpenAI-compatible option.

On Activepieces Cloud, the pricing FAQ says "On Plus and up you can connect your own AI provider keys, which drops every AI step to 1 credit and moves the model cost to your provider bill instead of ours." The AI Providers page does not say which edition it applies to, and we could not confirm on the pages we opened whether the free self-hosted Community edition includes it.

n8n

In n8n, each AI node uses a credential. The OpenAI credentials page lists one method, an API key: sign in to or create an OpenAI account, create a secret key, and add it as the API Key in n8n. An Organization ID is needed only if you belong to more than one organization.

On n8n Cloud Starter and Pro, Gateway credits let you run AI models from a prepaid balance without a key, and you can select Use my own credential to use your own account instead. That page says Gateway credits "aren't available on n8n Cloud Enterprise or self-hosted n8n." For n8n Assistant, the pricing page says that on self-hosted n8n "Assistant is available in preview" and "You bring your own key (BYOK), so no credit limits apply."

Which one fits your server

These are examples of common situations, with made-up people, to show how the facts above play out.

• Ana runs IT for a 40-person firm and wants automations for her own company, with SSO. Both licenses cover internal use, and both put SSO in a paid plan: Activepieces lists it from Team on Cloud and in Enterprise control, and n8n in its self-hosted Business and Enterprise plans.
• Ben is a developer who wants to change the code and ship his version to clients. The licenses differ most here, so he reads Activepieces' MIT terms and n8n's distribution limit side by side, and asks both vendors before he writes a line.
• Chloe wants to try without running a server. Activepieces Cloud Free needs no card, and n8n's 14-day trial needs no credit card either.
• Dev runs flows with an AI step in most of them. On Activepieces Cloud, he counts AI credits by model and compares that with Plus and his own keys. On a self-hosted n8n server, his AI nodes use credentials he creates, so his AI provider bills each call.

For a wider look at what automation tools charge per run, our guide to Zapier alternatives for solo freelancers counts three example workflows on several meters, and human-in-the-loop AI automation tools compares how each tool pauses for approval.

Where Barside fits

Barside runs as a desktop app on your own computer instead of on a server, on Windows, Linux and Mac. It has no app-to-app connectors or webhooks, so a job that starts in a web form or another web app still belongs in Activepieces or n8n. What it does is repeat AI work in your notes and, on Windows and Mac, in other apps on your computer. Saved workflows keep a job the AI assistant did once and run it again by name or on a schedule, and every scheduled run waits in Agents & approvals → Inbox until you press Approve. AI agents write morning briefings, call follow-ups and research notes, and anything they propose waits in the same inbox.

Both use the AI you connect in Settings → AI: Claude, ChatGPT or Gemini with an API key, OpenRouter with a one-click sign-in, or a Claude or ChatGPT plan through Claude Code or Codex with My plan. Barside charges nothing for AI, and an AI spending limit in the same place caps what each computer spends. Our guide to AI apps where you bring your own key covers the setup.

Know the limits before you count on it. Scheduled runs need your computer on and Barside open, and only the latest missed run from the last 12 hours is made up. With Barside Plus (US$20 a month or US$200 a year), cloud agents run scheduled agents on Barside's server, while saved workflows stay on your computer. You can keep up to 100 saved workflows with 40 recorded steps each. Workflows are stored on your computer, and when one runs, the request and the notes it reads go to the AI company you connected. Read how agents use your data.

Works well with

Barside runs on Windows, Linux and Mac, and you can download it free.