App hosting platforms

App hosting platforms deploy web apps, APIs, workers, and static sites without you managing servers by hand. This grid compares modern PaaS hosts and hyperscaler app services on minimum price, how far one app can scale, uptime SLAs, stability signals, what they can run, and which OS or runtime models they support. The real trade-off is simple usage-based PaaS versus hyperscaler depth, and long-running containers versus serverless or static-first platforms.

Last updated

How to choose

Start with what you are shipping. Long-running APIs and workers favor container PaaS (Render, Railway, Fly.io, Heroku, Northflank, Koyeb, DigitalOcean, Beanstalk, App Service, App Engine). Front-end and serverless-first apps often fit Vercel or Netlify better. Check whose metal you land on (own vs AWS/GCP/Azure), and whether Cloudflare sits in the edge path—helpful for global traffic, awkward if you already proxy through another CDN.

Rule out deal-breakers

  • Confirm the runtime, language, or container image you need is supported.
  • Decide whether free tiers that sleep or pause are acceptable.
  • Check whether you need a published uptime SLA on the plan you will buy.
  • Cap scale: some entry plans allow only one instance.

Compare the key trade-offs

  • Price shape: flat minimums, usage credits, or pay only for cloud resources under a free control plane.
  • Scale model: fixed replicas versus function concurrency versus autoscaling groups.
  • Ops surface: git-push PaaS versus cloud consoles wired into IAM and networking.
  • Company and platform risk: ownership changes and status history matter when the host is critical path.

Pitfalls

  • Free always-on is rare; many free services sleep or pause at limits.
  • Elastic Beanstalk has no platform fee, but EC2 and load balancers still bill.
  • Vercel and Netlify concurrency is not the same as VM replica counts.
  • Highest SLA tiers are often Enterprise-only.

Answer

Browse the table. Filter to what matters.

Use the filters below, then sign in if you want to shortlist and decide with friends.

Sign in to vote

Sign in to vote, hide services and features, and reorder.

Verified: hover a cell and click its green dot for the source Estimate: no cited source Not stated: the company doesn't publish it

FeatureService ⠿
Links
⠿
Links
⠿
Links
⠿
Links
⠿ ⠿
Links
⠿ ⠿
Links
⠿
Links
⠿
Links
⠿ ⠿
⠿
Price
$0/month, Free static; web from $7
$5/month, Hobby minimum usage
$2.19/month, shared-cpu-1x 256mb
$5/month, Eco dyno
$0/month, Sandbox free credits
$0/month, PAYG compute from $0.0022/h
$0/month, Free static; paid from $5
$0/month, Hobby
$0/month, Free forever
$0/month, Standard free daily hours
$0/month, No platform fee; pay AWS resources
$0/month, F1 Free; Basic B1 ~$13
⠿
Free tier
Yes
Yes
No
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
⠿
Cheapest paid server
$7/month, Less than 1 CPU 512MB
$5/month, Hobby monthly usage credit floor
$2.19/month, shared-cpu-1x 256MB
$5/month, Eco dyno 0.5 GB
$12/month, nf-compute-50 0.5 vCPU 1GB
$1.61/month, eco-nano 0.1 vCPU 256MB
$5/month, Paid basic container
$20/month, Pro plan + $20 credit
$9/month, Personal plan
$36.50/month, ~F1 always-on after free quota
$8/month, ~t3.micro only; ALB extra
$13/month, Basic B1 Linux approx
⠿
Max servers in cluster
100 servers, instances on paid up to 8c-32g
50 servers, Enterprise replicas; Pro 42
Not stated
255 servers, Fir dynos default
Not stated
Not stated
Not stated
30,000 servers, concurrent functions Hobby/Pro
Not stated
Not stated
Not stated
Not stated
⠿
Uptime SLA
Contractual SLAs on Enterprise (no figure published)
99.999%, Enterprise target; Pro 99.99
99.9%, Enterprise support
Not stated
99.99%, Enterprise
99.99%, Enterprise; Scale 99.9
99.95%, App Platform ACI
99.99%, Enterprise
99.99%, Enterprise
99.95%, covered configs
Not stated
99.99%, paid apps
⠿
Historic system stability
Public status page; recent all systems operational
Public status page linked; no uptime history reviewed
Public status site for Fly.io
Long-running PaaS with public status
Status not quoted on pricing pages
Public product status not quoted here
SLA-backed App Platform; DO status culture
Public status; Enterprise SLA
Public status; Enterprise SLA
Google Cloud status and SLA history
Depends on underlying AWS region health
Azure status and paid SLA
⠿
Company stability
Independent (Render Inc.)
Independent (Railway Corp.)
Independent Fly.io
Salesforce-owned PaaS
UK company Northflank Ltd
Joining Mistral AI
Public DigitalOcean product
Independent Vercel Inc.
Independent Netlify
Google Cloud product
Amazon Web Services product
Microsoft Azure product
⠿
Own infrastructure
No
Yes
Yes
No
No
Yes
Yes
No
No
Yes
Yes
Yes
⠿
Underlying cloud
AWS and GCP
Own metal; burst AWS/GCP
Own metal in colo DCs
AWS
GCP managed; BYOC multi-cloud
Leased bare metal + Cloudflare edge
DigitalOcean cloud
AWS primary; also Azure GCP
AWS; multi-cloud CDN history
Google Cloud
AWS (EC2 and related)
Microsoft Azure
⠿
Cloudflare powered
Yes
No
No
No
No
Yes
No
No
No
No
No
No
⠿
Workloads
Web services, static, workers, cron, Postgres
Apps, Docker services, open-source databases
Framework apps, agents, sandboxes, Postgres
Web and worker dynos; Salesforce-connected apps
Containers, jobs, databases, GPUs, BYOC
Containers, AI inference, GPUs, scale-to-zero APIs
Static sites and containerized apps
Web apps, AI apps, marketing sites, functions
Marketing sites, web apps, functions
Server-side web apps and backends
Web apps on EC2; lift-and-shift
Web and API apps on Azure
⠿
OS and runtime model
Debian 12 buildpacks or Docker
Docker images and Dockerfiles
Fast-booting VMs (Fly Machines)
Managed container dynos (Cedar/Fir)
Containers on Northflank Cloud or BYOC K8s
Secure VMs running containers
Managed containers from registries
Serverless functions and edge runtime
Static CDN plus serverless functions
Standard or Flexible environments
EC2 platforms; Linux or Windows IIS
Linux or Windows App Service plans
⠿
Languages and frameworks
Node, Bun, Python, Ruby, Go, Rust, Elixir, Docker
Any via Docker or git build
Rails, Phoenix, Django, Node, Laravel, Go, Rust, .NET
Buildpack and container languages
Any container image
Core runtimes plus any Docker image
Apps and containers from git or registry
Next.js first; Node and edge functions
Static generators, web apps, functions
Node, Java, Ruby, C#, Go, Python, PHP
Java, .NET, Node, Python, PHP, Ruby, Go, Docker
Common web stacks on Linux or Windows
⠿
Autoscaling
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
⠿
Sleeps or scale to zero
Free web sleeps after 15 min idle
No sleep called out on Hobby pricing
Machines can auto-stop when idle
Eco sleeps after 30 min inactivity
Sandbox always on, no sleeping
Scale-to-zero with sub-250ms cold start
Free static only; paid always-on containers
Functions scale; Pro cold-start prevention
Site pauses when monthly credits end
Standard can scale to zero
Typical always-on EC2; no free sleep tier
Free F1 CPU quota; paid Always On available
⠿
Regions
US and EU/Asia region list
Global regions; concurrent on Pro
Many cities worldwide
Up to ten global regions (Private Spaces)
17+ global regions
7 regions; Enterprise 50 on-demand
Several regions in North America, Europe and Asia
Global edge; multi-region compute
Global CDN edge
Google Cloud regions
All commercial AWS Regions
Most Azure regions
⠿
Custom domains
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
⠿
Managed database
Render Postgres and Key Value
Deploy open-source databases on Railway
Managed Postgres; Upstash Redis
Heroku Postgres and add-ons
Postgres, MySQL, MongoDB, Redis
Serverless Postgres
Dev and production databases; Postgres for dev
Bring your own or marketplace
Netlify Database (Postgres)
Google Cloud databases (Cloud SQL etc.)
AWS RDS and other AWS data services
Azure SQL and Azure Database for PostgreSQL

Why these features matter

  • Price ($/month)

    Starting cost filters hobby projects from paid platforms. Usage-based hosts can still spike after the floor.

  • Free tier

    Lets you prototype before committing. Free often means sleep, pause, or no SLA.

  • Cheapest paid server ($/month)

    Free tiers sleep or cap hard; this is the real floor to run a tiny always-on app.

  • Max servers in cluster (servers)

    Hard caps force plan upgrades or a ticket to support when traffic grows.

  • Uptime SLA (%)

    SLA credits are limited, but they show what the vendor will commit to on paid plans.

  • Historic system stability

    Past and current status culture is a weak but useful reliability signal.

  • Company stability

    Acquisitions and parent clouds change roadmap and pricing risk.

  • Own infrastructure

    Own metal can cut cost and egress surprises, but you inherit that maker's capacity and ops; hyperscaler layers inherit AWS/GCP/Azure outages and pricing.

  • Underlying cloud

    Data residency, outage blast radius, and egress/network behavior follow the substrate, not just the PaaS brand.

  • Cloudflare powered

    Cloudflare edge helps global latency and DDoS, but stacks another CDN, can break some PaaS features when double-proxied, and couples uptime to Cloudflare incidents.

  • Workloads

    Picking a static/serverless host for a long-running worker creates friction.

  • OS and runtime model

    Windows-only apps and custom kernels need the right host model.

  • Languages and frameworks

    Native runtimes are smoother than bringing your own image for every stack.

  • Autoscaling

    Autoscaling cuts overnight ops during traffic spikes.

  • Sleeps or scale to zero

    Cold starts and sleeping dynos break always-on APIs and cron assumptions.

  • Regions

    Latency and data residency follow region choice.

  • Custom domains

    Production brands need their own domain early.

  • Managed database

    One-vendor app plus DB simplifies ops; multi-vendor can be cheaper or safer.

About the services

  • Render · Render

    Git-push PaaS for web services, static sites, workers, cron, and managed Postgres. Strong free static hosting and a low paid web-service floor; free web services sleep after idle traffic.

  • Railway · Railway

    Usage-based cloud for apps and open-source databases with Docker and git deploys. Hobby starts at a $5 monthly usage minimum; replica limits rise sharply on Pro.

  • Fly.io · Fly.io

    Hardware-virtualized containers (Fly Machines) close to users worldwide, with optional Managed Postgres. No ongoing free tier; smallest shared CPU machines start around a few dollars a month if left running.

  • Heroku · Heroku

    Classic dyno PaaS with add-ons and Heroku Postgres, now under Salesforce. Eco dynos are the cheapest entry and sleep when idle; Fir generation raises default dyno caps.

  • Northflank · Northflank

    Container and Kubernetes-style PaaS with GPUs, BYOC, and a free Sandbox that stays always on within a resource credit. UK company; production plans are pay-as-you-go compute.

  • Koyeb · Koyeb

    Serverless containers with scale-to-zero and fast cold starts, including GPU options. Joining Mistral AI; Pro starts at $29/month plus compute, with pay-as-you-go compute from fractions of a cent per hour.

  • App Platform · DigitalOcean

    DigitalOcean managed apps with free static sites and paid containers from $5/month, plus a published 99.95% App Platform SLA. Fits teams already on DigitalOcean networking and databases.

  • Vercel · Vercel

    Frontend and serverless platform built around Next.js, Vercel Functions, and global edge. Hobby is free; Pro adds usage credits. Scale is measured in function concurrency, not VM replicas.

  • Netlify · Netlify

    Jamstack and web-app host with Git deploys, Functions, and optional Netlify Database. Free forever with monthly credits; sites pause when credits run out until the next cycle.

  • App Engine · Google

    Google Cloud serverless app platform with Standard and Flexible environments and a daily free instance-hour allowance on Standard. Strong Google Cloud integration; 99.95% SLA on covered configurations.

  • Elastic Beanstalk · Amazon Web Services

    AWS application management layer over EC2, load balancers, and Auto Scaling with no Beanstalk platform fee. You pay AWS resources only; supports Java, .NET, Node, Python, PHP, Ruby, Go, Docker, and Windows IIS.

  • App Service · Microsoft

    Azure managed web and API hosting on Linux or Windows plans, from Free F1 trials to Basic and Premium. Production usually starts on Basic; Microsoft publishes a 99.99% App Service SLA on paid plans.

Suggest a change

You’ll sign in or create a free account to send it. We’ll keep what you typed.

Related comparisons