Short answer first: the three platforms charge in three different shapes. Render adds a workspace fee for teams on top of fixed instance prices. Railway has no free tier and meters everything per second on top of a small plan fee. Fly.io is pure pay-as-you-go per second, with no base plan fee at all. The right pick depends less on the headline price and more on which shape fits how you work.
Prices are accurate as of June 2026. All three change pricing often — check the live page before you commit. The shapes change far less than the numbers.
The three shapes in one line each
- Render — pick an instance size at a fixed monthly price, plus a flat workspace fee for teams. Predictable, with a real free tier that sleeps.
- Railway — no free tier. A small monthly plan fee, then per-second metering of CPU, RAM, volumes, and egress on top.
- Fly.io — no free tier, no seat fee. You pay per second for each running machine and for what it uses. The most "raw cloud" of the three.
Is there a free tier?
| Platform | Free tier? | The catch |
|---|---|---|
| Render | Yes | Free web services spin down after ~15 min idle, then cold-start. Fine for demos, not production. |
| Railway | No | One-time $5 trial credit, no card. Then you are on a paid plan. |
| Fly.io | No | New accounts have no free allowance. You pay from the first running machine. |
How usage metering works
Render bills fixed instance prices — choose a size (Starter ~$7/mo for 0.5 CPU/512 MB, Standard ~$25/mo for 1 CPU/2 GB) and pay it whether the app is busy or idle.
Railway and Fly.io both bill per second of actual use:
- Railway meters memory, CPU, volumes, and egress on top of your plan fee. An always-on 1 GB service runs in the low tens of dollars/month past the included credit.
-
Fly.io bills each machine by the second. The smallest
shared-cpu-1x/ 256 MB is ~$2.02/mo if left on; more RAM is ~$5/GB/mo; egress in NA/EU is $0.02/GB. Stopped machines cost only their disk.
Per-second billing is cheaper for spiky or scale-to-zero workloads and harder to predict, because the bill moves with traffic.
Side-by-side
| Factor | Render | Railway | Fly.io |
|---|---|---|---|
| Free tier | Yes (sleeps ~15 min) | No ($5 trial credit) | No (new accounts) |
| Billing model | Fixed instance price | Per-second usage + plan | Per-second usage |
| Workspace / team fee | $25/mo flat (Pro) | $20/mo flat (Pro) | None |
| Entry cost (solo) | Free or ~$7/mo | ~$5–15/mo | ~$2–7/mo |
| Best for spiky traffic | Weak (fixed size) | Good | Good |
| Cost predictability | High | Medium | Low–medium |
When each wins
- Render — predictable, fixed bills you can budget; a real free tier for demos (if you accept cold starts); simple over tunable.
- Railway — clean DX with per-second billing that scales down when idle; great for spiky traffic; small enough team that the plan fee stays small.
- Fly.io — lowest raw compute cost, no seat fees, multi-region or scale-to-zero, and you are fine with more configuration.
The pattern behind all three
All three start cheap and grow with usage, plan tier, or both. That is the normal SaaS shape — convenience now, a bill that climbs as you succeed. Worth paying if you use what you pay for.
If the part you dislike is the climbing bill, that is what we are building Hostim to avoid: deploy a container or git repo, pay a flat predictable price for the resources you pick. New projects get a free 5-day trial (no card), and managed Postgres/MySQL/Redis have a free tier of their own.
We're building Hostim.dev to make this simpler — happy to answer any questions.
Top comments (0)