Free n8n Hosting in 2026: Where to Run n8n for $0

n8n itself costs nothing to run. The server it runs on is the bill, and most "free" hosting quietly breaks n8n.

n8n is free software when you host it yourself: n8n's own documentation says that "Without a license key, n8n runs as the free Community edition" - n8n docs (opens the program details). What a founder pays for is the machine and the database underneath it, and that is where the free options split in two. Some free tiers keep n8n running around the clock with its data intact. Others put the server to sleep after a few minutes, or give it no disk that survives a restart, and n8n on those tiers misses its scheduled runs or forgets its workflows.

This guide sorts every free route we could verify on the host's own page on November 13, 2026: n8n's own Cloud trial, app platforms with a ready n8n template, and cloud servers with an always-free machine. It ends with the cheapest paid hosts for when free stops being enough, and n8n's startup discount. Every price links to the page that states it.

The short answer

These are the free ways to run n8n today. They trade effort for permanence, and only some of them keep n8n awake with its data intact:

Sources for each row are in the sections below. The pattern is simple: n8n needs a server that stays up and a database that stays put. The trial is the fastest way to try it, Northflank is the easiest free place to keep it, and a free cloud server is the most generous if you can run Linux.

What n8n needs from a host

n8n is not a static website. It is a long-running server that does three things a free tier has to support:

It runs on a clock. A workflow with a schedule trigger fires at its time only if n8n is awake at that time. A host that puts idle services to sleep (Render spins down a free web service that "goes 15 minutes without receiving any inbound traffic" - Render (opens the program details)) wakes only when a request arrives, so a 6 a.m. schedule on a sleeping instance simply does not run.

It keeps state. Your workflows, saved credentials and execution history live in n8n's database. Railway's n8n template spells out the shape: it "runs the official n8n Docker image with Railway’s managed Postgres", and "n8n requires a server with at least 1 vCPU, 512MB–1GB RAM, and Postgres for persistence" - Railway (opens the program details). If that database sits on a disk that is wiped on restart, every restart is a fresh, empty n8n.

It listens for webhooks. Many workflows start when another service calls n8n's webhook URL. A sleeping instance answers the first call slowly or not in time, and some senders give up and retry later.

So the questions to ask of any free host are: does it sleep, does it keep a disk or a database across restarts, and is there enough memory for n8n (Railway's guidance puts the floor at 512 MB to 1 GB). Render's own n8n guide goes further for production and recommends "running your web service on the 1c-2g compute plan or larger" - Render (opens the program details).

One more thing to settle before you host it: what you are allowed to do with it. n8n's Community edition is free under the Sustainable Use License, which lets you "use or modify the software only for your own internal business purposes or for non-commercial or personal use" - n8n (opens the program details). Automating your own startup's work is fine. Reselling n8n as a hosted service, or building it into a product your customers use, is outside that free license.

Option 1: n8n Cloud's free trial

The quickest way to use n8n without hosting anything is n8n Cloud, where n8n runs the server. The Starter and Pro Cloud plans on n8n's pricing page (opens the program details) both start with a free trial, and the page says plainly that "No credit card is required for the n8n Cloud trial."

The trial is generous with features and tight on volume. n8n lists its "Trial limits: 1,000 executions, 5 concurrent executions, a 180-second execution timeout, and 1 team project", with Pro-level features such as insights and workflow history switched on, and 800 credits for n8n's AI Assistant that "expire when the trial ends" - n8n (opens the program details). That is plenty to build and test real workflows; it is not a free home for a workflow that runs every five minutes.

After the trial, n8n Cloud is paid. The Starter plan is 20€ a month billed annually for 2,500 workflow executions, and Pro is 50€ a month billed annually for 10,000 - n8n (opens the program details). Pricing counts executions "regardless of complexity", so a workflow with thirty steps costs the same as one with three.

Use it for: trying n8n, or a light production setup where 20€ a month is cheaper than your time spent running a server.

Option 2: Northflank's free Sandbox

Of the app platforms that deploy n8n from a template, Northflank's free tier is the one built the way n8n needs. Its Sandbox plan is "Free", "No card needed", gives "Up to $20/mo of resources on Northflank Cloud" across "2 services, 2 jobs, and 1 database", and is "Always on, no sleeping" - Northflank (opens the program details).

Northflank publishes an n8n template: "With Northflank, you can deploy n8n in minutes using the n8n stack template", which pairs the n8n service with "A PostgreSQL addon for storing workflows, credentials, and runtime data" - Northflank (opens the program details). One service and one database is exactly what the Sandbox allows, so n8n and its Postgres fit inside the free plan, with one service left over.

The limit to watch is the $20 of monthly resources. Northflank bills per second past it, "From $0.01667 per vCPU hour" - Northflank (opens the program details), so size the n8n service modestly and watch the usage page in your first week.

Use it for: a free, always-on n8n with a real database and no server to maintain. Our Northflank free plan page tracks the Sandbox terms.

Option 3: a free cloud server (Oracle, Google Cloud, Azure, AWS)

If you are comfortable on a Linux command line, the most generous free n8n is your own small server on a big cloud's free tier. n8n runs there the standard way, with Docker, and n8n's documentation has setup guides for several providers (its Hetzner guide (opens the program details) is one). You do the maintenance: updates, backups, a domain and HTTPS for webhooks.

Oracle Cloud has the largest standing free allowance. Every account gets "the first 1,500 OCPU hours and 9,000 GB hours per month for free" on Arm-based Ampere A1 VMs, which Oracle says "is equivalent to 2 OCPUs and 12 GB of memory", plus up to two small AMD VMs - Oracle docs (opens the program details). New accounts also get a US$300 credit for the first 30 days - Oracle (opens the program details). That is far more memory than n8n needs, free with no end date. Oracle asks for a card for "account setup and identity verification" - Oracle (opens the program details), and charges only "for services that exceed the monthly free amounts".

Google Cloud's Free Tier includes "1 non-preemptible e2-micro VM instance per month in one of the following US regions", alongside a $300 Welcome credit valid for 90 days - Google Cloud (opens the program details). The e2-micro is Google's smallest machine, so keep the workflows light. Google also documents running n8n on Cloud Run - Google Cloud (opens the program details), but Cloud Run is built "to run stateless containers" - Google Cloud (opens the program details), so n8n there needs a separate database that the Cloud Run free tier does not include.

Microsoft Azure's free account includes "750 hours each of B1s, B2pts v2 (Arm-based), and B2ats v2 (AMD-based) burstable VMs" for 12 months, plus a $200 credit for 30 days - Microsoft Azure (opens the program details). 750 hours covers one VM running all month, so n8n runs free for a year, then the VM is billed.

AWS now gives new customers "up to $200 over 6 months" on its Free plan - AWS (opens the program details), which ends "6 months from the date you opened your AWS account" or once the credits run out - AWS FAQ (opens the program details). It is a good way to run n8n on EC2 for half a year, not a permanent home.

Use it for: a free n8n you control end to end. Oracle is the one that stays free; Azure is free for a year, AWS for six months.

Option 4: Render and Railway, and why their free tiers fit n8n badly

Render and Railway are two of the most popular places to deploy apps, both have official n8n templates, and both are free to start. Their free tiers are designed for apps that can sleep or stay small, which is the wrong shape for n8n.

Render walks through "deploying n8n on Render using the official n8n Docker image" - Render (opens the program details). Its free tier gives 750 instance hours a month, but a free web service spins down after 15 minutes idle, and "Paid services can preserve local filesystem changes by attaching a persistent disk, but Free web services cannot." The free Postgres does not last either: "Free Render Postgres databases expire 30 days after creation" - Render (opens the program details). A free n8n on Render therefore loses its schedules to sleep and its database after a month. It is fine for a test; for anything you depend on, Render's paid instances with a disk are the route.

Railway's n8n template is the cleanest setup we found: n8n plus managed Postgres in one click - Railway (opens the program details). The Free plan starts "with a 30-day free trial with $5 credits, then $1 per month", needs no credit card, and caps each service at "1 vCPU / 0.5 GB RAM" - Railway (opens the program details). The trial covers a month of n8n; $1 a month afterwards does not. Railway bills memory at "$ 0.00000386 per GB / sec", which is about $5 a month for half a gigabyte held all month (our arithmetic), before CPU, storage and the database. Railway's Hobby plan "Includes $5 of monthly usage credits" and is billed by what n8n actually uses past that, which makes it one of the cheapest paid homes for n8n.

Other one-click n8n hosts

A few more platforms deploy n8n from a template. Their free allowances suit testing better than running n8n for good:

• Zeabur offers an n8n template "with automatic volume persistence" - Zeabur (opens the program details), so workflows survive restarts. Its Free plan is for "personal exploration and learning", and "Services on the Free Plan automatically sleep after a period of inactivity"; the Dev plan is US$5 a month after a 14-day free trial - Zeabur (opens the program details).
• Koyeb has an n8n one-click app - Koyeb (opens the program details). Its free instance scales to zero when idle, which pauses scheduled workflows the same way Render's does.
• DigitalOcean's Marketplace lists an "n8n 1-Click App" maintained by a third-party vendor - DigitalOcean (opens the program details). New accounts get a $5 signup credit that expires after 90 days, and Droplets start at $4 a month - DigitalOcean (opens the program details).

The cheapest paid n8n hosting

When a workflow matters enough that it must not sleep, miss a run or hit a trial cap, a few dollars a month buys a managed or self-managed n8n. These are the entry prices each host states for running n8n, cheapest first:

Sources: PikaPods, DigitalOcean (opens the program details), Railway (opens the program details), Hetzner (opens the program details) and its starting credit (opens the program details), Sliplane, Elestio (opens the program details) and its trial (opens the program details), n8n (opens the program details).

The managed options (PikaPods, Sliplane, Elestio, n8n Cloud) handle updates and backups for you; the server options (DigitalOcean, Hetzner) are cheaper per gigabyte of memory but leave the upkeep to you. Railway sits between them: you pick the template, Railway runs the containers and the database, and the bill follows usage.

One cost the table cannot show is executions. n8n Cloud prices by workflow executions, so a busy workflow can push you up a plan, while every self-hosted option above runs unlimited executions on the Community edition and is limited only by the server's size.

n8n's startup plan

n8n runs a discount for young companies on its top self-hosted tier. Its Startup plan is 333€ a month billed annually for 40,000 workflow executions, against 667€ for the regular Business plan, and adds SSO, SAML and LDAP, version control, environments and admin roles - n8n (opens the program details). To qualify, a startup needs "Fewer than 20 employees" and "Less than €5M in total funding", and the discount lasts "for one year, or longer if you still meet the requirements". The Business plan runs only on your own deployment, so you still need one of the hosts above.

Most early teams do not need Business features; the free Community edition covers the automations themselves. The startup plan matters once you need single sign-on, separate staging and production environments, or Git-based version control for workflows. Our n8n Start-up Plan page tracks its terms.

Which option to pick

You want to try n8n this afternoon: start the n8n Cloud trial (opens the program details). No card, nothing to deploy, 1,000 executions to learn with.

You want n8n free for good, without running a server: deploy Northflank's n8n stack on the free Sandbox. It does not sleep and keeps a real Postgres database.

You want n8n free for good and can run Linux: take an Oracle Cloud Always Free Arm VM and run n8n with Docker. It has the most memory of any free option; Google Cloud's e2-micro works for light use.

You want it reliable for a few dollars: PikaPods from $3.75 a month if you want it managed, Railway's Hobby plan if you want n8n and Postgres from one template, or a $4 Droplet if you want the server yourself.

You want n8n to run your business's core processes: budget for a paid instance with backups, whether n8n Cloud's Starter plan or a paid server, and look at n8n's startup plan once you need its Business features.

Whatever you choose, back up your workflows: export them, or keep n8n's database on a host that takes backups, because a free tier that changes its terms can take your automations with it. For the rest of your stack, our guides to free app hosting, free Postgres hosting and free AI agent builders compare what each host and tool gives away free.

Hosting free tiers change often, and several have ended in recent years. Every term here was read on the host's own page on November 13, 2026; confirm it on the linked page before you move workflows people rely on.

FAQ

Is n8n free? Yes, when you host it yourself: n8n's Community edition is free to self-host under its Sustainable Use License, for your own internal business purposes or personal use. n8n Cloud, where n8n hosts it for you, is paid after a free trial, from 20€ a month billed annually.

Can I host n8n for free forever? Yes, on an always-free tier that does not sleep and keeps a database: Northflank's Sandbox (up to $20 a month of resources, no card) or an Oracle Cloud Always Free VM. Tiers that sleep, such as Render's and Zeabur's free plans, miss scheduled workflows.

Does n8n Cloud have a free plan? No. It has a free trial with no credit card, limited to 1,000 executions, 5 concurrent executions and 1 team project. After the trial, the cheapest Cloud plan is Starter at 20€ a month billed annually.

What is the cheapest way to host n8n? Free on Northflank's Sandbox or an Oracle Always Free VM. Among paid hosts, PikaPods states managed n8n from $3.75 a month, DigitalOcean Droplets start at $4 a month, and Railway's Hobby plan starts at $5 a month with $5 of usage included.

How much memory does n8n need? Railway's n8n template states that "n8n requires a server with at least 1 vCPU, 512MB–1GB RAM, and Postgres for persistence". Render recommends its 1 CPU, 2 GB plan or larger for production n8n.

Can I use free n8n in a product I sell? Not under the free license. The Sustainable Use License allows use "only for your own internal business purposes or for non-commercial or personal use", so offering n8n to your customers needs a commercial agreement with n8n.