Railway vs Render: Which Is Cheaper in 2026?
Railway and Render both deploy a backend from a git push with no server to manage. They bill in opposite shapes: Railway meters every second of CPU and RAM, Render sells fixed instances. This page works out which shape is cheaper for your workload from the prices each vendor publishes. We build InstaPods, which appears once below as a third option.
The short answer
Render is cheaper when a service works all month, Railway when it mostly waits. Render Starter is a fixed $7/mo for 0.5 CPU and 512 MB. Railway bills about $20.29 per vCPU and $10.14 per GB of RAM over a 730-hour month, so a service holding 512 MB costs more than $7 there once it averages about a tenth of a vCPU; below that Railway is the cheaper one, at about $5 to $7. At full load, 1 vCPU and 1 GB used around the clock is about $30.43 on Railway against $25 for Render Standard (1 CPU, 2 GB).
All prices read off each vendor's own pricing page on 2026-10-09.Per-second metering, no instance sizes to pick, excellent dashboard.
Dimensions where the two are close enough that it should not decide anything.
Fixed instance sizes, fixed prices, no meter to watch.
Railway vs Render, row by row
A winner is called on every row. Where they are genuinely level, the row says so.
| Dimension | Railway | Render | Winner |
|---|---|---|---|
| Billing model | Per second, per vCPU used and per GB of RAM held | Fixed monthly instance price plus a workspace plan fee | Tie |
| Free tier | One-time $5 trial credit that expires in 30 days, then a Free plan with $1 of usage credit a month, capped at 1 vCPU / 0.5 GB per service | Free web service with 0.1 CPU / 512 MB and 750 instance hours a month; spins down after 15 minutes idle, about 1 minute to wake | Render |
| Cheapest paid plan | Hobby $5/mo, with $5 of usage included | Starter instance $7/mo (0.5 CPU / 512 MB) on a $0 Hobby workspace, never sleeps | Railway |
| Small service holding 512 MB all month | About $5.07 of RAM plus $20.29 per vCPU used: passes $7 once it averages about 0.1 vCPU | $7 flat on Starter whatever it uses, up to 0.5 CPU | Tie |
| Cost of 1 vCPU + 1 GB used 24/7 | About $30.43/mo at $0.0278/vCPU-hr and $0.0139/GB-hr | $25/mo Standard, which is 1 CPU and 2 GB | Render |
| Cost when mostly idle | CPU billed for the vCPU time the service actually uses, RAM for what it holds while it runs; the floor is the $5 Hobby plan | Full instance price regardless of traffic | Railway |
| Team plan fee | Pro $20/mo per workspace, counted toward usage, unlimited seats | Pro workspace $25/mo flat plus compute, unlimited team members | Railway |
| Managed Postgres | Metered like any other service, no fixed floor | $6/mo for 0.1 CPU / 256 MB, $19/mo for 0.5 CPU / 1 GB; 1 GB of storage included, then $0.30/GB | Tie |
| Persistent storage | Volumes $0.15/GB per month, up to 5 GB on Hobby | Disks $0.25/GB per month, paid services only | Railway |
| Bandwidth | Egress $0.05/GB from the first GB: 20 GB is $1.00, 100 GB is $5.00 | 5 GB included on Hobby, 25 GB on Pro, 1 TB on Scale, then $0.15/GB: on Hobby 20 GB is $2.25; on Pro 20 GB is $0 and 100 GB is $11.25 | Tie |
| Shell access | railway ssh in the CLI gives a shell, SFTP and local port forwarding; VS Code Remote-SSH is not supported | SSH on paid web services, private services and background workers; not on Free instances | Tie |
| Bill predictability | Usage-based, a traffic spike moves the invoice | Fixed instance price; only bandwidth, disk and database storage vary | Render |
Where Railway wins
You never size an instance
Railway has no instance types. Your service uses what it uses and you are billed per second: $0.00000772 per vCPU-second and $0.00000386 per GB-second. Nothing to overprovision, nothing to resize when traffic changes.
Cheaper when the service mostly waits
Railway bills CPU for the vCPU time a service actually uses and RAM for what it holds while it runs. A service that holds 256 MB and barely touches the CPU comes to about $2.54 of RAM a month, inside the $5 Hobby plan. On Render the same service is $7 on Starter whether it served a million requests or none, and the Free instance spins down after 15 minutes idle.
Team plan that counts toward usage
Railway Pro is $20 a month per workspace with unlimited seats, and that $20 is included usage. Render Pro is a flat $25 a month with unlimited team members, and compute is billed on top of it.
Where Render wins
A $7/mo price that stays $7
Render Starter (plan ID 0.5c-512mb on the pricing page) is $7 a month for 0.5 CPU and 512 MB on a $0 Hobby workspace, and it costs the same in a quiet month and a busy one. Railway has no fixed-price equivalent: the $5 Hobby plan includes $5 of usage and everything above that is metered.
Cheaper once the service is busy
Railway works out to about $20.29 a month for one vCPU used around the clock and $10.14 per GB of RAM. A service that uses 1 vCPU and 1 GB all month is about $30.43 on Railway. Render Standard (plan ID 1c-2g) is $25 for 1 CPU and 2 GB, and filling that size on Railway would be about $40.58.
Managed Postgres from $6/mo
Render Postgres starts at $6 a month for 0.1 CPU and 256 MB (Basic-256mb), and the 0.5 CPU / 1 GB size (Basic-1gb) is $19, with 1 GB of storage included and more at $0.30 per GB a month. On Railway a database is another metered service, so its cost tracks what it consumes.
What it actually costs
A Node API that uses about 1 vCPU and 1 GB of RAM around the clock, plus a small Postgres and 20 GB of egress
- Hobby plan: $5/mo, counted toward usage
- CPU, 1 vCPU used 24/7: about $20.29/mo
- RAM, 1 GB held 24/7: about $10.14/mo
- Postgres: metered on top
- Egress 20 GB: $1.00
- Hobby workspace: $0/mo, one seat
- Standard instance (1 CPU, 2 GB): $25/mo
- Postgres 0.5 CPU / 1 GB (Basic-1gb): $19/mo, 1 GB of storage included
- More database storage: $0.30/GB
- Bandwidth: 5 GB included, 15 GB over at $0.15/GB: $2.25
- Build plan: $7/mo (2 vCPU, 2 GB, 25 GB disk)
- PostgreSQL: installed in the same pod, $0
- Bandwidth: not metered
- SSH and cron: included
Railway numbers are computed from its published per-second rates at 730 hours, for a service that really uses a full vCPU. If the same API averages a quarter of a vCPU, the Railway compute lines drop to about $15.22 (0.25 vCPU plus 1 GB). Render numbers are list prices for the named instance and database sizes. The InstaPods column assumes you run Postgres inside the pod rather than buying a managed database, which is the trade: you get the price, you also get the responsibility for tuning and restoring it.
The third option: stop paying per service
Both platforms price each moving part separately. A web service, a worker and a database are three line items on Render and three metered services on Railway. InstaPods sells one real Linux server per pod at a flat price, and everything you want to run goes inside it: the app, the database, the cron jobs, the queue worker. The $7/mo Build plan is 2 vCPU, 2 GB of RAM and 25 GB of disk, which covers all of the above for most small production apps.
Honest limitation: there is no separate managed database product and no autoscaling. If your Postgres falls over at 2am, you are the one restoring it, and if traffic triples you resize the plan yourself rather than the platform absorbing it. Render Postgres at $19/mo exists precisely so you do not have to do that.
Deploy on InstaPods $3/mo flat - $10 credit when you add a cardHow to choose
Frequently asked questions
Is Railway or Render cheaper?
Render is cheaper for a service that is busy all month, Railway for one that mostly waits. Render Starter is $7/mo fixed for 0.5 CPU and 512 MB. On Railway a service holding 512 MB costs about $5.07 a month in RAM plus $20.29 per vCPU it actually uses, so it passes $7 once it averages about a tenth of a vCPU. At full load, 1 vCPU and 1 GB around the clock is about $30.43 on Railway against $25 for Render Standard (1 CPU, 2 GB). Prices checked 2026-10-09.
Is Railway or Render better?
Neither is better across the board. Railway suits prototypes and services with uneven traffic: there is no instance size to choose, CPU is billed per second, and egress is $0.05/GB. Render suits steady production services: a fixed monthly price per instance, a managed Postgres from $6/mo, and 5 GB to 1 TB of bandwidth included depending on the workspace plan. If you need a bill you can state in advance, pick Render; if you need to pay only for what runs, pick Railway.
Railway vs Render free tier: what do you get?
Render gives each workspace 750 Free instance hours a month on a 0.1 CPU / 512 MB web service that spins down after 15 minutes without inbound traffic and takes about one minute to come back. A Free Render Postgres database expires 30 days after creation. Railway gives a one-time $5 trial credit that expires in 30 days, then a Free plan with $1 of usage credit a month, capped at 1 vCPU and 0.5 GB per service. Paid Render instances from $7/mo do not sleep.
What does Railway actually charge per hour?
Railway publishes per-second rates: $0.00000772 per vCPU-second and $0.00000386 per GB-second, which works out to $0.0278 per vCPU-hour and $0.0139 per GB-hour. Over a 730-hour month that is about $20.29 per continuously used vCPU and $10.14 per GB of RAM. Railway quotes the same rates as $20 per vCPU and $10 per GB a month. Volumes are $0.15/GB a month and egress is $0.05/GB.
Does Render charge a platform fee on top of compute?
Only above the free workspace. The Hobby workspace is $0/mo plus compute with one seat, so a solo developer pays just the $7 instance. Pro is a flat $25/mo plus compute and includes unlimited team members and 25 GB of bandwidth. Scale is $499/mo with 1 TB of bandwidth. Render announced these flat plans on 2026-04-23 to replace per-member pricing. Railway Pro is $20/mo per workspace, and that $20 is counted toward usage.
Can I SSH into my app on Railway or Render?
Yes on both, with limits. Railway has a railway ssh command in its CLI that opens a shell in the running service and supports SFTP and local port forwarding; VS Code Remote-SSH is not supported. Render offers SSH on paid web services, private services and background workers, and not on Free instances. On both you get a shell inside one running service, not a server you administer.
Sources
Every number on this page was read from these pages on 2026-10-09. Vendors change pricing without notice, so check the source before you commit budget.