Skip to main content

Comparison · updated 2026-08-06

PaaS With Flat, Predictable Pricing

Seven container platforms, ranked by how little the invoice moves rather than by how good the product is. Every price below comes from the vendor's own published list and is linked at the end of each entry.

Every platform here can deploy a Dockerfile and attach a managed database. The thing that separates them is what happens to the invoice when your app has a busy month. Metered platforms charge more; reserved platforms charge the same and run slower. Neither is wrong — they fail in different directions, and the right one depends on whether your workload sleeps.

The order below is by billing predictability alone. It is not a quality ranking: the platforms at the bottom are excellent products with pricing models built for a different shape of workload.

1.

Hostim

Fully flat · no meter

You reserve a container size and pay that number every month. There is no CPU meter, no memory meter, no egress line and no build-minute charge, so next month's invoice equals this month's invoice. The trade is that you size the box yourself: undersize it and you get a slow app rather than a bigger bill.

Billing modelFlat monthly price per reserved container, billed hourly
Entry app€2.50/month for 1 vCPU and 1 GB RAM (plan sa-1-1), custom domain and HTTPS included
4 vCPU / 8 GB€13.50/month (plan sa-4-8)
Outbound trafficNo per-GB charge, no ingress fee
Build timeNot metered
DatabasePostgreSQL 256 MB, MySQL 256 MB and Redis 128 MB permanently free; €5/month for 5 GB

Where it is strong

  • Nothing on the invoice is usage-derived, so the bill can be forecast a year ahead.
  • The database free tier is permanent and uncapped in instance count, not a 30-day trial.
  • PostgreSQL and MySQL run with a hot standby and automatic failover on every plan, including free.

Where it falls short

  • One region, in Falkenstein, Germany. No North American or Asian region today, and no edge network.
  • No object storage, CDN, managed DNS or transactional email — those stay with an external provider.
  • Small and founder-led, in production since October 2025. Less track record and support scale than the larger names here.
  1. Hostim.dev price listHostim.dev (checked 2026-08-06)
  2. Pricing model — plan-based billing, no meteringHostim.dev docs (checked 2026-08-06)
2.

DigitalOcean App Platform

Fixed container · metered extras

The container itself is a fixed monthly price, which puts it near the top of this list. The rest of the stack is where the variability lives — bandwidth beyond the included allowance, and a managed database that costs three times the container.

Billing modelFixed price per container, plus metered bandwidth overage
Entry appFrom $5/month per container
Static sites$3/month each beyond the first
Outbound trafficIncluded allowance, then per-GB overage
DatabaseManaged PostgreSQL from $15/month; no free tier

Where it is strong

  • Container price is a fixed line you can plan around.
  • A full cloud sits around the app — Droplets, Spaces object storage, load balancers, managed Kubernetes.
  • Wide region list, including North America and Asia.

Where it falls short

  • The managed database costs more than the app it serves.
  • Bandwidth overage is still a usage line, so a traffic spike still moves the bill.
  • US company, so EU-region data remains reachable under the US CLOUD Act.
  1. App Platform pricingDigitalOcean (checked 2026-08-06)
  2. Managed databases pricingDigitalOcean (checked 2026-08-06)
3.

Render

Fixed instance · metered egress and disk

Instance prices are fixed, which is the part people like, and everything around them is metered — egress beyond the included allowance, disks per GB, database storage per GB. Predictable in the normal month, less so in the month something goes viral.

Billing modelFixed instance price, plus metered egress and storage
Entry appStarter web service $7/month for 0.5 CPU and 512 MB; free instance available with cold starts
Outbound traffic5 GB/month included on Hobby, then $0.15/GB
Disks$0.25/GB-month
DatabaseBasic-256mb $6/month, Basic-1gb $19/month, plus $0.30/GB-month storage. Free database expires after 30 days

Where it is strong

  • A genuinely usable free tier for prototypes, if you accept that free services spin down when idle.
  • Preview environments, cron jobs, autoscaling and a managed key-value store in one product.
  • Instance pricing does not track request volume.

Where it falls short

  • Egress at $0.15/GB after 5 GB is the most expensive traffic rate in this list.
  • Database high availability starts at Pro-4gb, $55/month — the Basic tier has no failover.
  • Workspace plans changed on 1 August 2026, so older cost comparisons no longer hold.
  1. Render pricingRender (checked 2026-08-06)
  2. Render vs Railway vs Fly.io pricingHostim.dev blog, June 2026 (checked 2026-08-06)
4.

Heroku

Fixed dyno · paid add-ons

A dyno is a fixed monthly price, so the compute line is predictable. The bill grows through add-ons instead — every capability that is included elsewhere is a separate paid subscription here, and the free plans are gone.

Billing modelFixed price per dyno, plus per-add-on subscriptions
Entry appEco dynos $5/month for 1,000 shared compute hours
Idle behaviourEco dynos sleep after 30 minutes of inactivity
DatabasePostgres Essential-0 from $5/month; free plans removed on 28 November 2022
InfrastructureRuns on AWS; EU region is Ireland

Where it is strong

  • The dyno price itself does not move with traffic.
  • The add-on marketplace and buildpacks remain the fastest path for a team that does not want to write a Dockerfile.
  • Fifteen years of documentation and answered questions.

Where it falls short

  • Add-ons turn one predictable line into several, and they are where the cost actually accumulates.
  • Eco dynos sleeping after 30 minutes rules them out for anything that must stay warm.
  • No free tier since November 2022, so the floor is a real bill from day one.
  1. Heroku pricingHeroku (checked 2026-08-06)
  2. Removal of Heroku free product plans FAQHeroku (checked 2026-08-06)
5.

Sevalla

Usage-metered · 25 locations

A Kinsta product with the widest reach in this list and a fully usage-based bill. Pods are priced per size but charged by the second, with outbound traffic and build time metered on top, so three lines move independently every month.

Billing modelUsage-based — pod seconds, egress and build time all metered
Entry appH1 $5/month for 0.3 CPU and 0.3 GB, with no custom domain; S1 $10/month for 0.5 CPU and 1 GB
4 CPU / 8 GBS4 at $160/month
Outbound traffic$0.10/GB; ingress free
Build time$0.02/minute
DatabaseFrom $5/month for 0.25 CPU, 0.25 GB and 1 GB storage; no free plan published

Where it is strong

  • Twenty-five data centre locations, including Ashburn, Los Angeles, Montréal, Tokyo and Sydney.
  • Deploys from a Dockerfile, buildpacks or your own image, with persistent storage available.
  • Kinsta behind it, which is a longer operating history than most platforms this size.

Where it falls short

  • At the same CPU and memory it is roughly ten times the price of the cheapest entry here, after converting currencies.
  • Three metered lines — compute, egress and build minutes — make a forecast harder than a single plan fee.
  • The $5 pod cannot serve a custom domain, so the real entry price for a live app is $10.
  1. Application hosting pricingSevalla (checked 2026-08-07)
  2. Application pricing — usage-based billing, egress and build timeSevalla docs (checked 2026-08-07)
  3. Database hosting pricingSevalla (checked 2026-08-07)
6.

Railway

Fully metered · per second

One of the smoothest deploy experiences in the market, attached to one of the least predictable invoices. Everything is metered — CPU, memory, volumes and egress — on top of a plan fee that is charged before any usage at all.

Billing modelPlan fee plus per-second metering of every resource
Entry appHobby $5/month including $5 of credit; Pro $20/month per workspace including $20 of credit
Compute rates$20 per vCPU-month and $10 per GB of RAM per month, charged on top of the plan fee
Outbound traffic$0.05/GB
Volumes$0.15/GB-month
DatabaseJust another metered service at the same CPU, RAM and volume rates

Where it is strong

  • Per-second billing genuinely scales down for spiky or part-time workloads.
  • Templates and one-click services make the first hour faster than anywhere else here.
  • The developer experience is the benchmark others are measured against.

Where it falls short

  • A steady always-on service pays the full metered rate every hour of the month, which is the worst case for this model.
  • No free tier — only a one-time $5 trial credit.
  • The plan fee is charged before usage, so the floor is $5 or $20 and the ceiling is open.
  1. Railway pricingRailway (checked 2026-08-06)
  2. Usage-based pricing: why Railway and Render bills creep upHostim.dev blog (checked 2026-08-06)
7.

Fly.io

Fully metered · cheapest compute

The cheapest per-second compute in this list and an invoice made of many small lines. Machines stop when idle and then cost only their disk, which is excellent for intermittent work and irrelevant for a service that must stay warm.

Billing modelPer-second metering across compute, storage and egress
Entry appAbout $2.32/month for a shared-cpu-1x machine with 256 MB left running; about $6.79/month at 1 GB (Fly's own examples, sjc region)
Outbound traffic$0.02/GB in the EU and US
Volumes$0.15/GB-month
DatabaseManaged Postgres Basic (shared-2x, 1 GB) $38/month plus $0.28 per provisioned GB

Where it is strong

  • Machines stop when idle and cost only their disk afterwards.
  • Fine-grained region placement, with more locations than anything else in this list bar Sevalla.
  • Egress at $0.02/GB is the cheapest traffic rate here if your app serves files.

Where it falls short

  • Managed Postgres at $38/month for 1 GB of RAM ends the cheap story at the database.
  • No free allowance — billing starts with the first machine.
  • Many small invoice lines is the hardest shape to forecast, even when the total is low.
  1. Fly.io pricingFly.io (checked 2026-08-06)

How this list was made

Prices were read from each vendor's public pricing page on 2026-08-06 and are quoted in the vendor's own currency, without conversion. Where a platform publishes rates rather than plans, the rate is quoted as published rather than turned into a monthly estimate we would have to defend.

Ordering is by how much of the bill is usage-derived, from none to all of it. Every entry lists where the platform falls short, including ours — a comparison where the author wins every row is an advertisement, not a comparison.

Left out: hyperscaler container services (AWS App Runner, Cloud Run, Container Apps), because their billing is metered by design and comparing them here would only restate that; and self-managed options like Coolify or Dokku on a VPS, where the fixed price is the server rather than the platform. Those are covered in the alternatives hub.

Frequently asked questions

Which PaaS has the most predictable monthly bill?

Hostim, because nothing on the invoice is derived from usage: you reserve a container size and pay a flat monthly price, with no CPU meter, no per-GB egress line and no build-minute charge. DigitalOcean App Platform and Render come next — both price the instance itself as a fixed line but still meter bandwidth and storage around it. Railway, Fly.io and Sevalla meter everything, which makes them cheaper for intermittent work and harder to forecast for anything always-on.

Why do usage-based hosting bills creep up?

Because the things that grow are not the things you watch. Compute is the line people budget for, and the surprises come from egress, build minutes, volume storage and retries — each small, each metered separately, each growing with something you did not plan. A background job that loops, a crawler that finds your API, or a chatty deploy pipeline moves the invoice without any change in real traffic. A flat plan removes that class of surprise by moving the failure mode: undersize the container and the app gets slow, which is visible immediately, rather than expensive, which is visible on the first of the month.

Is flat pricing always cheaper than usage-based pricing?

No. Usage-based billing is genuinely cheaper for workloads that are idle most of the time — a staging environment, a nightly batch job, a demo that runs during a sales call. Fly.io machines stop when idle and then cost only their disk, and Railway bills per second. Flat pricing wins when the service has to stay running: an always-on API, a queue worker, a websocket server, an inference container holding a model in memory. Those pay the full metered rate every hour of the month, and that is exactly where a reserved plan is cheaper as well as steadier.

What does a 4 vCPU, 8 GB container cost across these platforms?

Hostim charges €13.50/month for 4 vCPU and 8 GB on plan sa-4-8. The same CPU and memory on Sevalla is the S4 pod at $160/month, before egress at $0.10/GB and build time at $0.02/minute. Railway publishes $20 per vCPU-month and $10 per GB of RAM per month, which puts a continuously running 4 vCPU / 8 GB service at roughly $160/month of metered compute on top of the plan fee. The gap between a reserved plan and a metered one is widest exactly where the workload never sleeps.

Do any of these have a permanently free database?

Hostim does — PostgreSQL 256 MB, MySQL 256 MB, Redis 128 MB and a 1 GB volume, permanent and with no cap on how many instances you create. Render's free Postgres expires after 30 days. Sevalla starts managed databases at $5/month and publishes no free plan. Heroku removed its free plans on 28 November 2022. Railway has no free tier at all, only a one-time $5 credit. This is usually where a cheap stack stops being cheap.

What is the catch with flat pricing?

You do the sizing. A metered platform absorbs a traffic spike by charging you more; a reserved plan absorbs it by getting slower until you move to a larger plan. That is a better trade for most small teams because it is visible and bounded, but it is a real trade and worth stating. The other catch on Hostim specifically is reach: one region in Germany, so a North American or Asian user base is served faster today by Sevalla, Fly.io or DigitalOcean regardless of what the invoice says.

Get Started with Hostim.dev

Deploy any Docker image, Git repo, or template instantly. Explore in a free 5-day trial project – no credit card required.