Railway's pricing model is usage-based: you pay for the RAM and CPU your service holds, per minute, whether or not it's serving traffic. There is no spin-down on the standard setup — a deployed container sits there holding its memory allocation around the clock.
That inverts the intuition people bring from spin-down platforms. On Render, idle means asleep means free-ish. On Railway, idle means 720 hours a month of RAM-minutes quietly metered against your credit. A 512 MB Python service you forgot about is the classic way the monthly credit disappears.
The lever is the serverless/sleep setting (opt-in per service) and honest memory sizing — a service that needs 200 MB but requests 1 GB bills 5x for the same idleness.
Deploy-and-forget belongs on hard-cap platforms; Railway rewards deploy-and-watch.
The current plan structure and where the credit goes: https://toolfreebie.com/railway-heroku-alternative/
Top comments (0)