n8n Cloud vs. Self-Hosted: Which Edition Fits Your Business?
Compare n8n Cloud, self-hosted Community, Business, and Enterprise options by setup, maintenance, features, usage, cost, and licensing.
By chris · Published · Updated · 13 min read
n8n in Practice · Part 1 of 4
The short answer: choose n8n Cloud when you want to build workflows without operating the platform. Choose self-hosted Community Edition when infrastructure control, private connectivity, or custom packages matter and your team can maintain the service. Business adds shared projects, environments, source control, and SSO to a self-hosted deployment. Enterprise addresses stricter governance, support, and deployment requirements and is available as either n8n Cloud or self-hosted.
The decision comes down to the systems you need to reach, the people building workflows, and who will maintain the service. Here is how the options compare, with example workloads and budgets to help you choose.
Information reviewed September 7, 2026
Prices, limits, and feature boundaries change. Confirm the final choice on n8n’s pricing page and in your proposed agreement before purchasing. All prices and example budgets in this article are in US dollars (USD). Listed subscription rates assume annual billing. Confirm taxes and final billing terms at checkout.
The six practical choices
“Cloud versus self-hosted” leaves out an important detail. Hosting location and license tier are separate decisions. Business is currently self-hosted only. Enterprise can be hosted by n8n or deployed in your environment.
| Option | Current starting point | Who runs the platform? | Good fit |
|---|---|---|---|
| Cloud Starter | $20/month billed annually, 2,500 executions, 5 concurrent executions | n8n | Individuals and small businesses with a modest number of straightforward workflows |
| Cloud Pro | $50/month billed annually, 10,000 executions, 20 concurrent executions; a higher Pro tier is also available | n8n | Small teams that need more capacity, projects, admin controls, workflow history, and execution search |
| Self-hosted Community | No n8n subscription fee; infrastructure and operating time are separate costs | Your team or service provider | Technical teams that need deployment control, private network access, or unverified/private nodes |
| Self-hosted Business | $800/month billed annually, 40,000 executions | Your team or service provider | Teams that need SSO, shared projects, and Git-based deployment controls |
| Cloud Enterprise | Custom quote and execution allowance | n8n | Organizations that want managed infrastructure plus enterprise governance and dedicated support |
| Self-hosted Enterprise | Custom quote and execution allowance | Your team or service provider | Organizations with strict infrastructure, network, data, security, or deployment requirements |
Eligible startups with fewer than 20 employees can apply for the self-hosted Startup plan, subject to n8n’s funding and other eligibility criteria. Its US pricing page currently displays $333 per month billed annually for 40,000 executions. Check the Startup offer and eligibility requirements before subscribing.
What one execution actually means
n8n prices its paid plans around production workflow executions. One complete run counts as one execution even when the workflow contains many nodes. That is easier to estimate than pricing every action, but it still requires careful workload mapping.
- A workflow scheduled once per day runs about 30 or 31 times per month.
- A workflow scheduled every five minutes runs roughly 8,600 to 8,900 times per month.
- A webhook workflow receiving 100 events each day runs about 3,000 times per month.
- A chatbot may run once for every incoming message, depending on its design. Twenty conversations with ten messages each can mean 200 executions rather than 20.
Execution allowance measures monthly volume. Concurrency measures how many eligible production runs can execute at once. On Cloud, excess runs queue until capacity becomes available. Slow AI calls and large file jobs can create a backlog even with a modest monthly total. Manual, sub-workflow, and error executions are excluded from the regular production concurrency limit; waiting and resumed workflows need separate testing. See n8n’s concurrency guidance.
Manual editor runs, sub-workflows called with Execute Sub-workflow, and designated error workflows do not add to the paid execution quota. A scheduled run counts whenever its Schedule Trigger fires. A polling trigger counts when it finds new data; empty polls do not count. A test triggered through a published webhook or schedule still counts, including on a development instance. These distinctions come from n8n’s execution accounting documentation.
Cloud: easier to begin and easier to operate
n8n Cloud operates the servers and database. The instance owner can choose the release track, update cadence, and maintenance window; n8n applies security and stability updates automatically. Your team still owns workflow logic, credentials, API changes, data handling, and recovery from failed business processes. Review workflows after platform upgrades. See Cloud update controls.
Starter is reasonable for a small deployment where one shared project and five concurrent executions are enough. Pro adds more shared projects, greater concurrency, admin roles, global variables, workflow history, and execution search. Enterprise Cloud extends the hosted model with features such as more projects, much higher concurrency, external secret-store integration, log streaming, longer retention options, and dedicated support with an SLA.
The tradeoff is infrastructure control. n8n’s current Cloud documentation says customers cannot customize the base URL, use a custom domain, choose the database backend, configure ports or proxies, install custom TLS settings, tune queue mode or worker count, or freely change execution timeouts and pruning behavior. Cloud concurrency is tied to the selected plan.
Verified community nodes can be installed through the supported Cloud node picker when enabled. Unverified npm packages and private custom nodes need self-hosting. You can still call an unsupported service from Cloud using HTTP Request if its API and authentication are suitable. npm publication and n8n verification are separate statuses. Check the installation documentation and our node catalog for the integration you need.
Self-hosting: more control and more responsibility
A basic n8n container can be running quickly. Production self-hosting adds the work that keeps workflows dependable after the first afternoon:
- A supported database, normally PostgreSQL, with tested backups and a restoration procedure
- A stable encryption key that is backed up separately from the database
- TLS, a reverse proxy, correct webhook URLs, and a documented DNS setup
- SMTP or another delivery path for account and operational email
- Execution retention and pruning appropriate for your data volume
- Monitoring for the application, database, disk, queues, webhooks, and failed executions
- An update schedule with release-note review, staging, and rollback
- Resource limits and capacity planning for bursty, long-running, or file-heavy workflows
- Task-runner isolation and restrictions around Code, shell, filesystem, community, and custom nodes
- An owner who responds when an upstream API changes at 2 a.m.
Community Edition has no n8n subscription fee or paid-plan execution quota. It supports the core workflow engine and queue mode, so adding workers does not by itself require Business. Hardware, database performance, and API limits still constrain capacity.
Its main tradeoff for teams is collaboration: Community lacks shared projects, normal workflow and credential sharing, SSO, and built-in Git environments. The owner can access other users’ workflows, but that does not replace a shared project with assigned roles. Free registration also unlocks folders, editor debugging, and custom execution data. See n8n’s edition comparison.
Self-hosted Business adds six shared projects, SAML or LDAP SSO, Git-based environments, and longer insights. Its value is easier collaboration and controlled deployment. It still requires an infrastructure owner and currently includes forum support rather than a dedicated support SLA.
Self-hosted Enterprise is worth evaluating when you need external secret stores, log streaming, extended retention, multi-main availability, or dedicated support. Confirm feature entitlements and response targets in the quote. A subscription alone does not make a deployment compliant or highly available.
Paid self-hosted plans also need license activation and renewal connectivity. n8n’s pricing FAQ describes a daily license-server check that reports production execution usage. For restricted networks or disconnected deployments, agree on a supported arrangement with n8n before buying.
Four example buying decisions
These are hypothetical planning examples, not client case studies or measured capacity claims. Employee count alone does not determine the right plan.
1. An owner-led company automating routine office work
Imagine an office sending daily summaries and receiving inquiry or order events. Five daily schedules in a 30-day month produce 150 executions. Another 20 webhook events on each of 22 working days add 440, for 590 production executions. Planning for about 750 leaves roughly 25% headroom. Manual editor testing does not consume that allowance.
Cloud Starter is usually the simplest answer when its project and concurrency limits fit. Community Edition becomes attractive when the workflows must reach private systems, require an unverified package, or the owner already has reliable infrastructure support. Running a server only to avoid a $20 subscription rarely saves money once maintenance time is counted.
2. A small operations team handling several thousand events
Imagine a distributor syncing order and stock events. One hundred webhook events per day in a 30-day month produce 3,000 executions. A nightly report adds 30, for 3,030 executions. A planning allowance of about 3,800 adds roughly 25% headroom. Check peak traffic as well as the monthly average.
Cloud Pro offers a comfortable managed starting point and better team features. A self-hosted Community deployment can fit a single technical owner who needs private network access or custom nodes. A team that must share workflows and credentials should also price a paid edition. The choice depends less on the server’s raw cost than on who will patch it, monitor it, and restore it.
3. A growing organization with development and production controls
Consider a 30-to-80-person company running 20,000 monthly executions. Several people build workflows, production changes need review, identity management requires SSO, and the company wants separate development and production instances connected through Git.
Those governance requirements point toward self-hosted Business even when a cheaper plan could handle the execution count. Shared projects and a controlled release process are the reasons to upgrade in this example. The organization still needs someone to operate every instance and coordinate compatible upgrades.
4. A regulated or mission-critical operation
A company may require centralized secrets, externally streamed audit logs, defined retention, contractual response times, private network placement, or tighter control over where data is processed.
Cloud Enterprise is the shorter operational path when n8n’s hosted architecture and contractual terms satisfy those requirements. Self-hosted Enterprise provides more deployment control when the environment must live inside the organization’s infrastructure. It also leaves more infrastructure responsibility with that organization.
A better way to compare total cost
Use the same worksheet for every option:
| Cost area | Cloud | Self-hosted |
|---|---|---|
| n8n subscription | Published plan or enterprise quote | None for Community; Business or Enterprise license for commercial features |
| Compute and database | Included in the hosted service | Server, database, storage, network, and possible Redis or worker infrastructure |
| Backups and recovery | Platform infrastructure managed by n8n; your workflow export and business continuity plan still matter | Your responsibility, including restoration tests and encryption-key recovery |
| Updates and monitoring | Platform managed by n8n; workflow and upstream API monitoring remain yours | Platform, database, operating system, containers, integrations, and workflows |
| Incident support | Forum on standard plans; dedicated SLA support on Enterprise | Your team or provider, plus n8n support according to the license tier |
Cloud often wins when internal engineering time is scarce. Community Edition can win when the organization already operates similar services or needs controls Cloud does not expose. Business and Enterprise make sense when their governance and support features solve requirements that would otherwise create risk or manual work.
Setup effort and a sample monthly budget
For planning, allow roughly 30–90 minutes to open a Cloud workspace and connect a straightforward service, or 30–60 minutes for a basic Docker demonstration on a prepared host. A small production self-hosted deployment may take 1–3 working days to configure, secure, document, and test for recovery. SSO, network approvals, high availability, and complex migrations can extend that substantially. These are working estimates, not installation benchmarks or service quotes.
For a modest self-hosted Community system, an illustrative budget could be $40/month for compute and storage + $15 for off-site backups and monitoring + 2 hours of maintenance at an assumed $100/hour = $255/month. Those dollar figures are budgeting inputs, not researched hosting prices or Black Swamp AI rates. Replace them with your provider quotes and actual labor cost. Business adds its license charge on top.
Two hours might cover a routine update, backup check, and log review in a quiet month. Allow extra time for incidents, integration changes, and recovery exercises. Cloud removes much of the infrastructure work, but workflow maintenance remains. Model APIs, SMS, email, and other connected services can generate separate bills on every edition.
How Black Swamp AI can help within n8n’s license
n8n describes itself as fair-code and source-available. Its Sustainable Use License permits internal business use and explicitly allows consulting services such as building workflows, developing closely related custom features, and installing or maintaining n8n on a customer’s internal server.
Our normal engagement model follows that path. The customer owns its n8n Cloud account or self-hosted instance and its credentials. We can help evaluate the edition, design workflows, deploy and secure a self-hosted environment, build private nodes for internal APIs, document recovery, and provide ongoing maintenance.
n8n’s licensing FAQ distinguishes consulting on a customer’s instance from hosting customer workflows and credentials in a provider’s own instance, which it says requires Enterprise. Exposing n8n’s workflow interface inside a product requires an Embed arrangement.
A customer-facing application does not automatically require Embed. The Sustainable Use License FAQ gives an allowed example of a chatbot powered by a company’s own credentials. It gives a disallowed example of collecting customers’ HubSpot credentials to access their data. The concrete use of n8n and whose credentials it handles matters.
We scope provider-hosted and embedded services with n8n before offering them. The right agreement depends on the delivery model; a standard Enterprise subscription should not be assumed to grant all resale rights. For a specific case, request written guidance from license@n8n.io. This is an explanation of published guidance, not legal advice.
Our recommendation process
We start with the workflows rather than a preferred plan. A useful assessment records:
- Each trigger and expected monthly frequency
- Peak concurrency and typical workflow duration
- Applications, credentials, network locations, and data sensitivity
- Built-in, verified, unverified, or private node requirements
- Who can build, approve, deploy, and inspect workflows
- Backup, retention, logging, support, and recovery expectations
- The person or provider responsible for routine maintenance and incidents
That produces a defensible Cloud or self-hosted recommendation and a realistic operating plan. It also exposes cases where the simplest edition is enough, which is valuable information before committing to infrastructure or a commercial contract.
Explore our automation services and published n8n integrations.
n8n in Practice · Part 1 of 4
This series follows n8n from the buying decision through AI agents, custom integrations, and production operations.
- Cloud vs. self-hosted (you are here)
- n8n as an AI agent harness
- Community nodes, private integrations, and reusable workflows (coming next)
- Security, scaling, monitoring, and ownership (coming next)
Sources and review notes
- n8n plans and pricing
- n8n Startup plan
- n8n Cloud features and configuration limits
- n8n self-hosted security audit
- n8n source control and environments
- n8n Sustainable Use License
- n8n licensing FAQ for consulting, hosting, and embedding
Feature, pricing, and license references reviewed September 7, 2026. Verify changing commercial details directly with n8n.
Researched and edited with GPT-6 Astra. Reviewed by Chris at Black Swamp AI.
Have a project in mind?
We can help scope a local AI build or automation workflow.
Discuss it with Chris ↗