Cheapest n8n VPS Hosting in 2026: Real Self-Hosting Cost

The cheap part of self-hosting n8n is renting the server. The expensive part is owning the failure modes: a bad update, a full disk, an exposed editor, or a backup that was never tested. A useful cost comparison therefore has to include renewal pricing and operations, not only the largest discount printed on a hosting page.

I re-checked the official n8n pricing page, n8n Docker documentation, GitHub releases API, and Hostinger India VPS page on 21 September 2026. The result is more precise than the old “about $7” answer: a 4 GB KVM 1 plan was advertised at ₹599/month equivalent, with a stated ₹999/month renewal. Hostinger also states that plans are paid upfront and that the monthly figure is the total plan price divided by its term. Treat ₹599 as a promotional monthly equivalent, not a month-to-month bill.

The short answer: n8n Cloud vs a 4 GB VPS

The currencies differ, so I will not hide an exchange-rate assumption inside a dramatic “X times cheaper” claim. The defensible conclusion is simpler: n8n Cloud charges for a managed execution allowance, while a VPS is a fixed infrastructure bill. Self-hosting becomes financially stronger as executions rise, but only if somebody is willing to own the server work.

The official n8n page now says all Cloud plans include unlimited users and workflows. Starter includes 2,500 monthly executions and 5 concurrent executions; Pro includes 10,000 monthly executions and 20 concurrent executions. That corrects an older version of this article that described active-workflow limits. Current Cloud pricing is based on complete workflow executions, not the number of steps inside each run.

What the VPS price really costs after renewal

The headline ₹599 number is not the whole budget. Hostinger's page showed these KVM 1 facts:

• 1 vCPU core
• 4 GB RAM
• 50 GB NVMe storage
• 4 TB bandwidth
• ₹599/month promotional equivalent
• ₹999/month renewal for two years
• payment upfront, with the displayed monthly rate derived from the plan total

Because the initial checkout term was not stated in the rendered pricing card I captured, I cannot honestly publish an initial invoice total. The useful calculation is an annualised run rate, clearly labelled as arithmetic rather than a quoted bill:

Promotional annualised run rate = ₹599 × 12 = ₹7,188
Renewal annualised run rate     = ₹999 × 12 = ₹11,988
Difference                      = ₹400 × 12 = ₹4,800 per year

That makes the promotional equivalent roughly 40% below the renewal rate: (999 - 599) / 999 = 40.04%. Put ₹999, not ₹599, into a long-term budget. Before buying, open the checkout and verify the selected term, taxes, and final upfront total because those are account- and checkout-state facts.

KVM 1 also includes weekly backups on the public plan page. I still would not count one provider-managed backup as a complete n8n recovery plan. You need a PostgreSQL dump, the n8n data volume or required instance data, and the encryption key stored away from that VPS. A provider snapshot can recover a server; it does not replace an application-level restore test.

Cost per 1,000 n8n executions

A fixed server has no commercial “execution pack,” but dividing its monthly equivalent by actual runs makes the economics easier to compare. This is original arithmetic you can reproduce with one Python command:

python - <<'PY'
for monthly, label in [(599, "VPS promo"), (999, "VPS renewal")]:
    for runs in (2500, 10000, 50000):
        print(label, runs, round(monthly / (runs / 1000), 2))
PY

These columns are deliberately left in their source currencies. They show the shape of each model, not a fake currency-neutral comparison. Cloud cost follows an allowance and managed feature set. VPS unit cost falls as volume rises, until CPU, memory, database I/O, or your maintenance time forces a larger plan.

Also remember what counts as an execution. A single workflow run with 30 nodes is one execution on the current n8n pricing page. A polling workflow scheduled every minute can create about 43,200 runs in a 30-day month even when it finds no useful work. Webhook triggers and less frequent schedules can change the bill far more than shortening a workflow.

Is 4 GB RAM enough for n8n?

Four gigabytes is a sensible starting size for one n8n instance, PostgreSQL, and a reverse proxy under light production use. It is not an official capacity guarantee. Workflow payload size, concurrency, Code nodes, file handling, and retained execution data matter more than the number of workflows visible in the editor.

Do not use a made-up “workflows per GB” ratio. Measure the live stack:

docker stats --no-stream
docker compose ps
docker compose logs --tail=100 n8n
free -h
df -h
docker system df

For a seven-day baseline, record memory during normal traffic and during the largest expected workflow. Check whether the kernel killed a process:

docker inspect -f '{{.State.OOMKilled}} {{.State.ExitCode}}' n8n-n8n-1
sudo dmesg -T | grep -i -E 'out of memory|killed process'

A 1 vCPU plan can also become the limit before RAM does. Large Code-node transforms, PDF work, compression, or many concurrent executions compete for that single core. Move to 8 GB or more when measurements show sustained pressure, not because a generic sizing table says every n8n server needs it.

The official Docker documentation recommends Docker for most self-hosting needs and currently points readers to Docker Compose as the recommended installation method. It uses SQLite by default and supports PostgreSQL. Even with PostgreSQL, n8n recommends persisting /home/node/.n8n because it contains important instance data beyond the SQLite database.

The real self-hosting bill includes operations

The n8n Community Edition can be self-hosted, but “free software” does not mean a zero-cost service. Add these jobs to the budget:

1. Updates. Pin a stable n8n image, read release notes, dump the database, then update intentionally. On 21 September 2026, GitHub's latest-release endpoint returned n8n@2.39.10 as stable. The releases feed also showed n8n@2.40.5 as a prerelease. Do not mistake the newest version number for the stable channel.
2. TLS and network exposure. Bind n8n to 127.0.0.1:5678 and publish only the HTTPS reverse proxy. Docker-published ports can bypass the UFW path people expect.
3. Backups. Store a PostgreSQL dump and encryption key off-server. Test that pg_restore --list can read the dump, then perform a real restore drill on a separate database.
4. Monitoring. Watch disk, memory, failed executions, container restarts, certificate renewal, and backup age. A workflow editor loading successfully does not prove scheduled jobs or webhooks work.
5. Incident time. Decide who responds when an API retry loop fills the database or an update changes a node's behaviour.

For the full deployment, use my n8n Docker Compose installation guide. It includes PostgreSQL, a localhost-only port binding, Caddy HTTPS, configuration validation, backups, and a safer update sequence. Keeping setup detail there avoids turning this cost page into a second thin version of the same guide.

Free n8n hosting options and their hidden constraints

Running n8n on hardware you already own can make the incremental hosting invoice zero. It still depends on your power, internet connection, router, storage, and ability to recover the machine. That can be reasonable for personal automations that may pause without causing damage.

Cloud free tiers change frequently, can sleep, limit outbound traffic, or reclaim idle capacity. I could not verify a universally available free VPS from the official sources used for this update, so I am not presenting one as the cheapest production recommendation. “Free forever” is a claim to verify at signup, not a foundation for a client workflow.

A laptop is useful for learning, but sleep and changing networks make it a poor webhook server. A Raspberry Pi can run n8n, yet buying one solely to avoid a VPS adds hardware, storage, and power costs before the first workflow runs. Existing hardware and new hardware are different cost cases.

Hostinger's VPS page listed n8n in its one-click application catalogue when checked. A template saves installation time; it does not remove ownership. Inspect the database choice, bound ports, encryption-key storage, update procedure, and backup path before putting credentials into any prebuilt image.

When self-hosting n8n is the wrong choice

Choose n8n Cloud when no one on the team owns Linux patching, restore tests, TLS, monitoring, and after-hours failures. The VPS can be cheaper on paper and still cost more than Cloud after one broken client automation consumes several hours.

Cloud is also the clearer choice when you need a managed support path or Cloud-only collaboration and governance features. Do not assume Community Edition contains every Business or Enterprise feature merely because the workflow engine runs without an execution allowance.

This single-VPS model is wrong for strict high availability. The server, database, reverse proxy, and local storage share one failure domain. Queue mode, workers, external storage, and database redundancy can address scale and resilience, but they also erase the simplicity that makes a small VPS cheap.

Finally, do not self-host regulated or sensitive workloads just to save money. Data location, retention, access logging, secret management, and incident obligations need an explicit review. Root access is control, not compliance.

My recommendation

Start with a 4 GB VPS if you already know how to maintain Linux or are willing to learn on a non-critical workflow. Budget at the ₹999/month stated renewal equivalent, even if checkout gives you the ₹599 promotional equivalent. Use PostgreSQL, keep port 5678 on localhost, preserve the encryption key, and send verified backups off the server.

For a business that cannot tolerate missed webhooks, price one hour of the operator's time beside the hosting bill. If that hour costs more than the managed premium and no one wants the responsibility, buy n8n Cloud. If you already operate Docker services and your workflows are execution-heavy, the fixed-cost VPS model is usually the cleaner economic fit.

FAQ

What is the cheapest reliable VPS for n8n?

A 4 GB VPS is the smallest configuration I would use as a default for n8n, PostgreSQL, and a reverse proxy. Hostinger India showed KVM 1 at ₹599/month promotional equivalent and ₹999/month renewal on 21 September 2026, but verify the upfront checkout total and term before buying.

Does self-hosting n8n cost money?

The Community Edition software can be self-hosted without a per-execution hosting fee, but the server, domain if needed, backups, monitoring, and operator time still cost money. Existing home hardware can reduce the incremental invoice to zero while adding power and reliability constraints.

Is n8n Cloud cheaper than self-hosting?

It can be when workflow volume is low and server maintenance would consume paid time. At the checked annual-billing state, Starter was €20/month for 2,500 executions and Pro was €50/month for 10,000; a VPS uses a fixed infrastructure price instead of an execution allowance.

How much RAM does self-hosted n8n need?

Start with 4 GB for a light production stack and measure docker stats, OOM state, disk use, payload size, and concurrency. Large files, Code nodes, and parallel runs can require more memory even when the workflow count is small.

Does self-hosted n8n have unlimited workflows and executions?

It does not use the Cloud plan's commercial monthly execution allowance, but the server is not physically unlimited. CPU, RAM, database I/O, disk capacity, execution retention, and concurrency become the real ceilings.

Is Hostinger's one-click n8n template enough for production?

It is enough to shorten deployment, not to prove production readiness. Verify HTTPS, localhost port binding, database persistence, encryption-key storage, backups, restore steps, update ownership, and monitoring before depending on the template.

Should I use n8n Cloud or a VPS?

Use Cloud when you want managed operations and a support path. Use a VPS when you can own Linux and backups, need a fixed-cost execution model, or plan to run several small services on the same infrastructure without making the server a dangerous single point of failure.

Read next

• VPS vs Shared Hosting: Which Should You Choose in 2026?
• Atmospheric Water Generators: The Physics, the Yield, and the Real Cost Per Litre
• What Off-Grid Power Actually Costs Per Month (2026 Breakdown)

Some links on this page are affiliate links. If you buy through them I earn a commission at no extra cost to you. I pay for the services I write about and only recommend what I actually use.