Render vs Railway vs Light Cloud

For a small SaaS with an API and Postgres, Light Cloud is the cheapest of the three at $5 a month, Railway comes to about $5.30 once the RAM and CPU it bills by are counted, and Render costs $13.30. On the free plans a side project can cost $0 on all three, with different catches. And for a service that has to stay on around the clock, the order flips: Railway's metered service is cheapest at about $5.60, Render's $7 instance is close behind, and Light Cloud is the most expensive at about $52.

We make Light Cloud, one of the three, so all three bills are below, including the one we lose.

How we ranked

Render and Railway get compared constantly, usually on features. I wanted the bill, and one bill isn't enough, because the three platforms charge in three different ways. Render sells fixed instances by the month. Railway meters the RAM and CPU a service holds, by the minute. Light Cloud bills instance-hours only while a container runs and scales it to zero in between.

So I priced three workloads from the vendors' own pages, checked on 2026-10-13 and again on 2026-11-02, in US dollars, counting what each workload uses on a metered plan rather than the plan's starting fee:

1. A side project on the free plans: a static frontend and a small API, no database, no custom domain.
2. The reference small SaaS from where to host a SaaS app: React frontend, one Node.js API, small Postgres, custom domain, about 1,000 API requests a day.
3. A 512 MB service that must never sleep, like an API that can't afford a cold start.

The ranking table uses the second, because it's the app most people searching "Render vs Railway" are trying to place. Both other bills come after the three entries.

Kept running all month, with no scale to zero, the small SaaS gets the Always on prices in the table. Light Cloud comes last there at about $56, because an always-on instance needs the $19 Starter plan and a Micro costs about $52 a month, while Render's $7 instance and Railway's metered service already run around the clock.

The ranking

Scale to zeroAlways on

#PlatformIdle previewsAlways-on serviceSmall SaaS a month

1

Light Cloud

Apps that sit idle most of the day, with a preview of every branch that costs nothing until someone opens it

$5

Apps that sit idle most of the day, with a preview of every branch that costs nothing until someone opens itPrice: all in, Lite plan

Idle previews$0, every branch, every planAlways-on serviceFrom Starter, about $52

$5

all in, Lite plan

2

Railway

Cron jobs, private networking and pull request environments that copy the database, for about 30 cents more than us

~$5.30

Cron jobs, private networking and pull request environments that copy the database, for about 30 cents more than usPrice: Hobby $5 + about $0.30 of usage

Idle previewsBilled as usage while openAlways-on serviceAbout $5.60, billed by RAM and CPU used

~$5.30

Hobby $5 + about $0.30 of usage

3

Render

Services that are busy all day, at fixed prices you can read off the pricing page

$13.30

Services that are busy all day, at fixed prices you can read off the pricing pagePrice: $38.30 with full-stack previews

Idle previewsFull stack on the $25 Pro workspaceAlways-on service$7 fixed, every plan

$13.30

$38.30 with full-stack previews

List prices from each vendor's own pricing and docs pages, checked 2026-10-13, US dollars. Ranked by the monthly bill for the reference app: a React frontend, one Node.js API, a small PostgreSQL database and a custom domain, about 1,000 API requests a day. Light Cloud usage uses the pricing page's assumption of 0.2 s of instance time per request. Railway is priced on the usage the app generates, never on the plan's starting fee: about 250 MB of RAM for the API, which also serves the built frontend, and 200 MB for Postgres on average, 0.03 vCPU, a 1 GB volume and 1.5 GB of egress. The other two bills, a free side project and a 512 MB service that runs around the clock, are in the post. The Always on view keeps the same API running all month with no scale to zero.

1. Light Cloud

Best for: apps that sit idle most of the day.

Lite is $5 a month and includes $5 of usage, a custom domain with free SSL on custom domains and one shared PostgreSQL database (pricing). The small SaaS uses about $4.63: $4 for the database and 30 cents for a GB of storage, about 20 cents for the API, a few cents of builds and two cents per preview. That's what scale-to-zero pricing buys: the API is billed for 1.7 instance-hours a month, not 720.

One thing neither competitor offers: any environment can sit behind a visitor password, on any plan, with no add-on. Railway also has previews on every plan; here they're free branch previews too, Free included, and an idle one costs $0. The platform also deploys Express, FastAPI or Spring Boot apps without a Dockerfile.

Against Render and Railway, it has the fewest building blocks. No background workers, no cron jobs, no private networking; services talk over public HTTPS. Shared databases run in Belgium only. Previews share the production database until you give them their own. A container waking from zero takes 3.1 to 3.7 seconds for its first response.

2. Railway

Best for: more built in, for about 30 cents more.

Hobby is $5 a month including $5 of usage (plans). Usage is RAM at $10 per GB a month and CPU at $20 per vCPU a month, metered by the minute, with free builds. For the small SaaS that's about $5.33: roughly $2.50 for a Node API averaging 250 MB that also serves the built frontend, $2 for Postgres at 200 MB, $0.60 of CPU, and cents for the volume and egress. The bill is the larger of $5 and the usage, so about $5.30, and the same always on; a hungrier app pays the difference. Railway recommends Pro, $20 a month, once the app is commercial.

It has more of the plumbing than we do: private networking between services, cron jobs at least five minutes apart, and pull request environments that mirror the production services, database included. A hard usage limit can be set from $10. Sleeping a service is opt-in and starts 5 to 10 minutes after its last outbound traffic, and Railway says the first request to a sleeping service may return a 502. Free-plan apps get no custom domain after the trial.

3. Render

Best for: services that are busy all day, at fixed prices.

A Starter web service is $7 a month for 512 MB, and the smallest Postgres is $6 plus $0.30 per GB (pricing): $13.30 for the small SaaS, with the frontend as a free static site. Prices are per service and don't move with traffic, which makes Render the easiest of the three to budget once an app is busy.

It also has the most complete toolbox: background workers, cron jobs (minimum $1 a month each), a private network, autoscaling on Pro. The cost is previews. Full-stack preview environments with databases need the Pro workspace at $25 a month flat, which makes the small SaaS $38.30 plus preview compute. The free Hobby workspace previews one service per pull request.

The other two bills

Same sources and date, two more workloads:

The side project is a tie on price, so the catches decide it. Render's free web service spins down after 15 minutes without traffic and takes "about one minute" to wake; its free Postgres expires after 30 days, which matters if you add one later. Railway's Free plan after the trial is $1 of credit a month, 0.5 GB of RAM per service, one project and no custom domain. Ours has $1 of usage, unlimited static sites, three server apps and a preview per branch, a "by Light Cloud" link on each page and no databases. When the dollar runs out, server apps pause and static sites keep serving; the Express tutorial runs on it.

The always-on service is where we lose, and it isn't close. On Light Cloud an always-on instance needs the Starter plan, and a Micro running all month is about $52. Railway bills what the service holds: 0.5 GB of RAM all month is $5, and a light API averaging 0.03 vCPU adds about $0.60, so about $5.60, the cheapest of the three. Render charges $7 flat for the same 512 MB, whatever the service actually uses. If your app is a worker that never stops, Render or Railway is the answer, and I'd say the same to a friend.

The smaller meters

Compute and databases decide most bills, but three smaller meters can surprise you, and here we don't come out best. Traffic served by our containers is $0.20 per GB (rate card); static sites have bandwidth included up to fair use, 1 TB a month on paid plans. Railway charges $0.05 per GB of egress, the cheapest of the three. Render's Pro workspace includes 25 GB and then charges $0.15 per GB. For an API that sends large responses, that line can outgrow the plan.

Builds go the other way for Railway: they're free there. Render includes 500 build minutes a month on Hobby, then $5 per 1,000. We charge $0.01 a minute from the first one, which is a cent for a typical static deploy and $10 if you ever burned through 1,000 minutes. Storage for a database is $0.30 per GB a month on both Render and Light Cloud.

Where the others beat us

Most of it is in the second table. Fixed monthly prices for busy services are Render's strength, and metered RAM and CPU make Railway cheap for steady small workloads. Both have cron jobs and private networking, and Render has background workers as a service type. Railway's pull request environments copy the database; ours share the production one until you change that. And both have been around longer than we have.

Where Light Cloud comes out ahead is the shape of a young product: idle most of the time, a handful of branches open, a staging site that needs a password, a database that has to exist from day one. On that shape the bill is $5, and a Railway alternative that charges nothing for idle previews is worth a look. If you've already read our Render alternatives ranking, this is the same arithmetic with Railway at the table.

Which of the three bills looks most like your app?