Self-hosted n8n can use the free Community edition, but operating it still costs money and time. When evaluating n8n self hosted pricing, separate the software edition from the server, backups, external services, and people responsible for keeping workflows dependable. A low server bill describes only one part of the arrangement.
For a small team, the useful question is not “Can we install it cheaply?” It is “Can we run the workflows we need, with the right features and a manageable recovery process, at a cost we understand?”
TL;DR
- Community is a free self-hosted edition; paid self-hosted plans unlock additional features.
- Hardware or VPS costs, backups, external API usage, and maintenance remain separate budget items.
- Estimate workflow volume and resource needs independently. More nodes do not automatically mean the same thing as more billable executions.
- Compare Cloud and self-hosted arrangements at an equivalent level of responsibility and required features.
- Use a pilot to measure memory, execution duration, data growth, and recovery effort before committing to a production budget.
First separate the edition from the hosting
n8n’s edition documentation distinguishes Community, Registered Community, Business, and Enterprise. Community does not require a paid software subscription; commercial plans provide a license key for their additional features. Exact feature availability can change, so match requirements against the current plan information.
This means “self-hosted” is a deployment choice, not the name of a single free plan. A team can operate its own infrastructure and still pay n8n for features it needs. Conversely, a small internal workflow may fit Community without a commercial subscription.
Write down requirements before comparing prices: who must edit workflows, how credentials are shared, whether centralized login is required, and what support or governance the organization expects. A server upgrade cannot add a software feature that belongs to a different edition.
The official pricing page, checked on November 5, 2026, describes paid allowances in monthly workflow executions rather than charging for every node in a workflow. Treat the current plan page and your agreement as the authority for allowances, billing periods, and feature limits; do not apply another automation tool’s billing unit to n8n.
The six parts of a self-hosted budget
| Budget component | What to include | What changes the cost |
|---|---|---|
| Software | Required n8n edition and applicable license | Collaboration, governance, support requirements |
| Runtime | Compute, memory, database, persistent storage | Concurrency, payload size, execution duration |
| Recovery | Backups, retention, restore work | Data volume and acceptable recovery time |
| External services | APIs, AI models, email, other connected tools | Calls, tokens, messages, provider plans |
| Operations | Updates, monitoring, troubleshooting | Change frequency and team experience |
| Business continuity | Coverage when the owner is unavailable | Importance of the workflows and support expectations |
A budget should identify which costs are cash payments and which are staff time. Combining them is useful for decision-making, but calling both a monthly hosting invoice creates confusion.
Runtime cost depends on the workload
A workflow that receives a few small webhooks is different from one that downloads large files, transforms them, and waits for several external services. Counting workflows alone does not describe memory use, storage growth, or peak concurrency.
Begin with the busiest normal hour rather than the average day. If several scheduled jobs all start at nine in the morning, a server may struggle even though the monthly execution count looks modest. Staggering schedules may help, but it should not hide a genuine capacity requirement.
Keep execution data in the budget. Saving large inputs and outputs can make diagnosis easier, yet it can also grow the database rapidly. Define retention around the evidence you need for support and the sensitivity of the data, rather than keeping every payload indefinitely.
For an ordinary API-based deployment, a Cloud VPS can supply the underlying runtime. Check current specifications and the needs of your workflows; there is no universal server size that makes every n8n installation production-ready.
A residential address or proxy is not part of the basic n8n cost model. Add a separate network service only when a permitted task actually requires it. Paying for an address type does not make workflow logic, credentials, or external API permissions more reliable.
External services can exceed the hosting bill
An automation platform coordinates work that other services may charge for. AI requests, document processing, email delivery, database storage, and paid application subscriptions can all contribute to the total.
Estimate usage using the provider’s actual billing unit. For example, an AI call might depend on input and output tokens, while another service charges per document or request. A workflow execution is not a universal unit that can be multiplied by one price across all providers.
Consider failure paths too. A retry can repeat an external request, and a badly placed loop can produce far more calls than expected. Put spending alerts or usage limits around services where unexpected activity would be expensive.
Separate optional enrichment from essential processing. If a model-generated summary fails, the underlying customer inquiry should not necessarily be lost. A workflow that preserves the basic result can remain useful while an optional service is unavailable.
A transparent example budget
The following is a hypothetical planning exercise in US dollars. These numbers are invented to demonstrate the method; they are not VoyraCloud prices, n8n quotes, or a measured customer’s results. The example assumes Community meets the team’s requirements.
| Monthly item | Illustrative assumption | Example amount |
|---|---|---|
| Server | One modest runtime | $18 |
| Backup storage | Retained recoverable copies | $7 |
| External API usage | A capped allowance for connected services | $25 |
| Maintenance time | Two hours valued at $40 per hour | $80 |
| Total modeled operating cost | Cash plus assigned labor value | $130 |
The cash items total $50. Including the assigned value of maintenance time produces $130. Both figures are useful, but they answer different questions. Neither includes initial setup, a paid n8n plan, unusual incidents, or every possible connected-service subscription.
Now change one assumption. If maintenance takes five hours instead of two, the modeled total becomes $250. If external API usage doubles while maintenance stays at two hours, the modeled total becomes $155. This illustrates why a server-only comparison can exaggerate savings.
Do not turn the example into a universal break-even point. A team with experienced administrators may need less routine effort; a team with complex workflows may need much more. Replace every line with observed usage or an actual quote before using the budget for a purchasing decision.
Estimate executions without confusing them with outcomes
For planning, a scheduled workflow running once per hour over a thirty-day month has 720 scheduled starts. A webhook workflow receiving 100 events per day has roughly 3,000 incoming events over the same period. These are simple workload estimates, not a statement about how every plan bills retries, sub-workflows, tests, or failed runs.
Check those details against the current plan and your actual execution records. One business event can trigger more than one workflow, and an execution can process more than one record. The relationship depends on how the workflow is designed.
Also distinguish volume from value. Ten thousand successful executions do not prove that ten thousand correct customer updates occurred. Track duplicates, missing outputs, and rejected results alongside infrastructure usage.
For budgeting, keep a short worksheet with the trigger frequency, typical payload size, expected duration, external calls, and failure behavior. This is more informative than listing the total number of nodes on the canvas.
What self-hosting makes you responsible for
Someone must maintain the operating environment, protect credentials, manage updates, monitor availability, and restore service when something breaks. An application image can simplify installation, but it does not remove these continuing responsibilities.
Plan a repeatable update process. Keep a recoverable copy, check changes relevant to your workflows, and validate important integrations after the update. A successful application restart does not prove that every credential and business workflow still works.
Backups need a restoration test. A folder full of files is not enough if the team cannot reconstruct the instance or recover the information needed to use its credentials. Record what is backed up, where it is stored, who can access it, and how recovery is performed.
Set an explicit retention policy for execution records and review it as data grows. Keep enough evidence to investigate failures, then remove records that no longer serve a defined purpose. Plan routine cleanup rather than waiting for storage to fill up.
If nobody can own this work, a hosted service may be the more economical choice even when its subscription looks higher than a VPS invoice.
Check licensing before offering automation to others
Using n8n internally and building a service around it for external customers can raise different licensing questions. Do not assume that visible source code means every commercial use is unrestricted. Review the official repository license and obtain clarification from n8n when the proposed arrangement is unclear.
This matters for agencies as well as software companies. Managing a client’s own environment, reselling access to a shared platform, and embedding workflow functionality in another product are not automatically equivalent arrangements.
Keep this decision separate from infrastructure selection. A cheaper host does not resolve a licensing requirement, and a paid server subscription is not a license to use every feature of the application installed on it.
When Cloud or self-hosting fits better
A hosted service can be attractive when the team wants to focus on workflows, lacks infrastructure coverage, or values a simpler operational arrangement. Self-hosting can fit when the team has a specific deployment requirement and the ability to maintain the environment.
Compare the same requirements on both sides. Include necessary features, expected usage, support expectations, backups, and the value of administrator time. Do not compare a fully supported commercial plan with a bare server and describe the difference as pure savings.
The VoyraCloud n8n page is a starting point for the infrastructure route. The existing n8n deployment guide covers setup context; it should follow the edition and ownership decision, rather than substitute for it.
A five-step budget check
- List the workflows, owners, and required collaboration features.
- Select an edition that actually meets those requirements.
- Estimate runtime, storage, external-service use, and recovery needs separately.
- Run a representative pilot and replace assumptions with observations.
- Review the total after the first normal operating period, including maintenance and exceptions.
Keep initial setup separate from recurring operations so the first month does not distort the long-term comparison. Also record what would trigger a redesign: sustained resource pressure, a new governance requirement, or an external-service bill that exceeds the agreed limit.
FAQ
Is self-hosted n8n free?
The Community software edition is free to use under its applicable terms. Hosting, backups, connected services, and maintenance are separate costs. Some requirements may call for a paid edition.
Does every node count as a paid execution?
Do not equate nodes with executions. n8n’s current pricing describes monthly workflow executions. Check the plan’s detailed rules for your particular workflow patterns and billing arrangement.
Is the cheapest VPS enough?
Not necessarily. Payload size, concurrent work, database growth, and recovery expectations matter. Start with a representative workload and observe resource use rather than choosing only by the lowest advertised price.
Should I budget for a residential proxy?
Only if a specific task requires that network component. Ordinary API integrations and internal business workflows do not inherently need residential egress.
Can I ignore staff time if we already employ an administrator?
You may not receive an additional invoice, but the time still has an opportunity cost. Showing cash and labor separately makes the decision clearer without pretending they are identical expenses.
What should I measure during a pilot?
Measure peak resource use, execution duration, data growth, external-service consumption, failures, and recovery effort. Those observations make n8n self hosted pricing a practical operating budget instead of a comparison between headline subscription numbers.