ngpaas.eu

Railway vs. Render: Pricing, Regions and Fit Compared

Railway vs. Render compared: usage-based against instance-based pricing, EU regions in Amsterdam and Frankfurt, free tiers, databases and which suits your app.

PaaSPublished

Render suits teams that want a fixed price per running instance, a mature set of service types and a region in Frankfurt. Railway suits teams that run many small or bursty services and would rather pay for the CPU and memory those services actually consume, with a European region in Amsterdam. Both deploy from Git or a Dockerfile, both offer managed Postgres, and both are US companies. The decisive difference is the pricing model, not the feature list.

Both platforms are popular Heroku alternatives, and both publish their own comparisons. This page is independent: it is based on the providers’ documentation, not on benchmarks we ran.

At a glance

As of October 2026. The providers’ own pricing pages are authoritative.

Railway Render
Pricing model Subscription with included usage, then per resource Per instance size, plus workspace plan
Entry plan Hobby: 5 USD/month incl. 5 USD usage Free instances; paid instances per size
Usage rates (per provider) RAM 10 USD/GB/month, CPU 20 USD/vCPU/month, egress 0.05 USD/GB, volumes 0.15 USD/GB/month Fixed per instance type
Regions California, Virginia, Amsterdam, Singapore Oregon, Ohio, Virginia, Frankfurt, Singapore
Free option One-time trial credit 750 free instance hours per workspace and month
Company seat USA USA

Sources: Railway plans, Railway regions, Render regions, Render free tier.

Pricing: usage against instances

Railway: pay for what runs

On Railway you pay a small subscription that already contains the same amount of usage credit. On the Hobby plan that is 5 USD, on Pro 20 USD. Beyond that, every service is billed for the memory and CPU it consumes over time, plus outbound traffic and volume storage. A service that uses 0.5 GB of RAM continuously costs roughly half of the RAM rate per month; one that mostly waits costs much less.

The upside is that preview environments, small workers and internal tools barely register on the bill. The downside is that a memory leak or a traffic spike shows up directly as cost. Set usage limits and alerts from day one.

Render: pay for what you reserve

Render charges a fixed monthly price per instance type for web services, workers and databases, plus a workspace plan for team features. You know the bill in advance; the instance costs the same whether it handles ten requests or ten thousand. If it runs out of memory, it restarts rather than charging more.

That predictability is why many agencies and small companies prefer Render for client projects. The catch: an instance that idles most of the day still costs its full price.

To compare both models with your own services, use the PaaS cost calculator.

Regions and data location

For European users, both offer one region on the continent. Railway’s EU West region is in Amsterdam; Render’s is in Frankfurt. For an audience in Germany, Austria or Switzerland the latency difference between the two cities is small.

More important is a detail in Render’s documentation: an existing service or database cannot be moved to another region. If you start in Oregon by default and later need the EU, you create new services and migrate data. Choose Frankfurt at creation time if you serve European customers.

Hosting in an EU region does not change which law the provider answers to. If that matters for your customers, read sovereign cloud explained and check the data processing agreement of each provider.

Developer experience

Railway is built around a project canvas: services, databases and their connections appear as a graph, and variables can reference each other. Spinning up a database next to an app takes seconds, and pull-request environments are straightforward. It is particularly pleasant for prototypes and for teams that run many small services.

Render organises resources in a more classic list with clear service types: web service, private service, background worker, cron job, static site, Postgres and a Redis-compatible key-value store. Infrastructure can be described in a render.yaml blueprint, which helps when you set up the same stack for several clients.

Databases and state

Both offer managed PostgreSQL. Render’s free Postgres database is limited to 1 GB and expires 30 days after creation, according to its documentation, so it is a test tool only. On Railway, databases run as services and are billed through the same usage model, with volumes for persistent data.

Whichever you choose, schedule your own regular dumps and store them outside the platform. That is your insurance against mistakes and your exit route.

Which one to choose

  • Choose Railway if you run many small services, preview environments or tools with irregular traffic, and you are willing to watch usage.
  • Choose Render if you want fixed monthly costs, clear service types, a Frankfurt region and infrastructure as code via blueprints.
  • Consider neither if company seat in the EU is a hard requirement; French platforms are listed in our Heroku alternatives comparison, and DigitalOcean App Platform is another instance-priced option with several European regions.

Providers in this comparisonAs of: October 2026

Frequently asked questions

Is Railway cheaper than Render?

It depends on load. For services that sit idle most of the day, paying for consumed resources on Railway is often cheaper. For services that run at steady load around the clock, a fixed Render instance is easier to predict and can be cheaper. Model both with your own numbers.

Does Render have a free tier?

Yes. According to Render, each workspace gets 750 free instance hours per month; free web services spin down after 15 minutes without traffic, and a free Postgres database expires 30 days after creation. It is meant for testing, not production.

Can I host in the EU with Railway and Render?

Yes. Railway offers an EU West region in Amsterdam and Render offers Frankfurt. Both providers are US companies, so check their data processing agreements and sub-processor lists if you handle personal data.

Which is easier to leave later?

Both build from a Dockerfile, so the application itself is portable. Plan the database exit early: regular dumps in standard PostgreSQL format make a later move straightforward on either platform.

More in PaaS