Hosting cost calculator
Pick an instance size and monthly traffic to compare the estimated cost of a single always-on service across platforms — including self-hosting on a VPS. Estimates, not quotes.
Estimated monthly cost of a single always-on service — a rough estimate, not a quote. Real bills vary with usage, add-ons and region; confirm on each provider. Each provider's included free bandwidth is subtracted before egress is billed. Sizes are normalized targets, so the mapped plan is the closest match, not an exact one — providers with no small plan (e.g. some VPS) are priced at their smallest available plan, which can over-provision the Small/Medium tiers. Open the on any row for the exact spec priced.
Shares your exact size + traffic.
Cheapest estimate: Hetzner Cloud at $7.09 per month; 14 platforms priced.
| Platform | Compute | Egress | Est. / mo | Verified |
|---|---|---|---|---|
Hetzner CloudⓘHetzner Cloud has NO micro/nano instances — the entire lineup starts at 2 vCPU / 4GB, so all three sizes map to the same entry plan, CX22 (2 shared Intel/AMD vCPU, 4GB RAM, 40GB NVMe) = $7.09/mo all-in (server + one primary IPv4). LARGE (2 vCPU / 4GB) is an exact match; SMALL (~0.5 vCPU/512MB-1GB) and MEDIUM (~1 vCPU/2GB) have no smaller/cheaper option so they use the CX22 floor (it over-provisions them). Prices are the 24/7 monthly cap, incl. one primary IPv4, excl. VAT, EU locations (FSN/NBG/HEL). Next tier up: CX32 (4 vCPU / 8GB) = $10.59/mo all-in. ARM alternative CAX11 (2 Ampere vCPU / 4GB) = $7.59/mo all-in. | $7.09 | — | $7.09 · lowest | 2026-07-16 |
Hetzner Cloud (self-host)ⓘHetzner Cloud has NO micro/nano instances — the entire lineup starts at 2 vCPU / 4GB, so all three sizes map to the same entry plan, CX22 (2 shared Intel/AMD vCPU, 4GB RAM, 40GB NVMe) = $7.09/mo all-in (server + one primary IPv4). LARGE (2 vCPU / 4GB) is an exact match; SMALL (~0.5 vCPU/512MB-1GB) and MEDIUM (~1 vCPU/2GB) have no smaller/cheaper option so they use the CX22 floor (it over-provisions them). Prices are the 24/7 monthly cap, incl. one primary IPv4, excl. VAT, EU locations (FSN/NBG/HEL). Next tier up: CX32 (4 vCPU / 8GB) = $10.59/mo all-in. ARM alternative CAX11 (2 Ampere vCPU / 4GB) = $7.59/mo all-in. | $7.09 | — | $7.09 · lowest | 2026-07-16 |
DigitalOcean DropletsⓘBasic Droplets (shared vCPU). SMALL = smallest Basic Droplet 512MB RAM / 1 vCPU @ $4/mo (DO has no 0.5-vCPU option; 1 vCPU is the shared-CPU minimum). MEDIUM = 2GB RAM / 1 vCPU @ $12/mo (exact match). LARGE = 4GB RAM / 2 vCPU @ $24/mo (exact match). Prices are per-second billed with a monthly cap, so flat monthly rate applies for 24/7 use. | $12 | — | $12 | 2026-07-16 |
VultrⓘRegular Cloud Compute (shared vCPU). SMALL = 1 vCPU / 1 GB / 25 GB @ $5/mo. MEDIUM = 1 vCPU / 2 GB / 55 GB @ $12/mo. LARGE = 2 vCPU / 4 GB / 80 GB @ $24/mo. Hourly-billed with a monthly cap; a cheaper IPv6-only $2.50 plan and High Frequency / High Performance tiers also exist. | $12 | — | $12 | 2026-07-31 |
Akamai (Linode)ⓘAkamai (Linode) Shared CPU. SMALL = Nanode 1 GB (1 vCPU / 1 GB / 25 GB) @ $5/mo. MEDIUM = Linode 2 GB (1 vCPU / 2 GB / 50 GB) @ $12/mo. LARGE = Linode 4 GB (2 vCPU / 4 GB / 80 GB) @ $24/mo. Hourly-billed with a monthly cap; Dedicated and High-Memory plans cost more. | $12 | — | $12 | 2026-07-31 |
DigitalOcean Droplet (self-host)ⓘBasic Droplets (shared vCPU). SMALL = smallest Basic Droplet 512MB RAM / 1 vCPU @ $4/mo (DO has no 0.5-vCPU option; 1 vCPU is the shared-CPU minimum). MEDIUM = 2GB RAM / 1 vCPU @ $12/mo (exact match). LARGE = 4GB RAM / 2 vCPU @ $24/mo (exact match). Prices are per-second billed with a monthly cap, so flat monthly rate applies for 24/7 use. | $12 | — | $12 | 2026-07-16 |
Fly.ioⓘFly Machines are usage-based (billed per-second), priced here as always-on ~730h/mo. Figures are the round Amsterdam (ams) baseline shown on the official pricing page, where the region markup ratio = 1.0 (ams is the base rate itself, not a marked-up region). Fly has no 0.5 vCPU preset; the smallest is shared-cpu-1x (1 shared/burstable vCPU). SMALL -> shared-cpu-1x + 1GB = $5.70/mo. MEDIUM -> shared-cpu-1x + 2GB = $10.70/mo. LARGE -> shared-cpu-2x + 4GB = $21.40/mo. These are shared/burst-CPU presets; dedicated-CPU 'performance' presets cost several times more. Effective cost is region-dependent: cheapest are Ashburn VA (iad) and Secaucus NJ (ewr) at ~0.75x the ams baseline; most expensive (Johannesburg jnb, Mumbai bom) at ~1.21x. | $10.7 | $2 | $12.7 | 2026-07-16 |
KoyebⓘSMALL = micro (0.5 vCPU / 512MB) or eco-small (0.5 vCPU / 1GB), both $5.36/mo. MEDIUM = eco-medium (1 vCPU / 2GB), an exact spec match, $10.71/mo (Standard "small" is 1 vCPU/1GB at the same $10.71). LARGE = eco-large (2 vCPU / 4GB), an exact spec match, $21.43/mo (Standard "medium" is 2 vCPU/2GB at the same $21.43). Koyeb's Standard and Eco price ladders share the same monthly figures ($5.36 / $10.71 / $21.43) at these tiers; Eco gives more RAM per vCPU. Monthly figures are Koyeb's own advertised Price/month (billed by the second; hourly = $0.0072 / $0.0144 / $0.0288). | $10.71 | $2 | $12.71 | 2026-07-16 |
CloudwaysⓘMapped to Cloudways Flexible on DigitalOcean, Basic/"Standard" (SSD) droplets (managed cloud VPS), verified live on the official pricing page 2026-07-16 (DigitalOcean provider, Basic server type, Standard toggle). SMALL -> Micro: 1GB RAM / 1 vCPU / 25GB SSD / 1TB transfer = $11/mo. MEDIUM -> Small: 2GB RAM / 1 vCPU / 50GB SSD / 2TB transfer = $24/mo. LARGE -> Medium (Most Popular): 4GB RAM / 2 vCPU / 80GB SSD / 4TB transfer = $46/mo. (1GB is the smallest plan; no 0.5-vCPU/512MB option exists.) A DigitalOcean "Premium" NVMe tier is also offered at roughly +$3/tier: Micro $14 / Small $28 / Medium $54 / Large $99. | $24 | — | $24 | 2026-07-16 |
DigitalOcean App PlatformⓘApp Platform container instances, billed as a flat monthly price for a 24/7 always-on service. SMALL = smallest shared "Basic" instance 1 vCPU / 512 MiB @ $5/mo (DO has no 0.5 vCPU option; this is the closest to the 512MB-1GB target). MEDIUM = shared 1 vCPU / 2 GiB @ $25/mo (exact match). LARGE = shared 2 vCPU / 4 GiB @ $50/mo (exact match). Intermediate tiers exist: 1 vCPU / 1 GiB fixed @ $10, 1 vCPU / 1 GiB shared @ $12. | $25 | — | $25 | 2026-07-16 |
NorthflankⓘSMALL = nf-compute-50 (0.5 shared vCPU / 1GB) $12/mo; MEDIUM = nf-compute-100-2 (1 dedicated vCPU / 2GB, "Most popular") $24/mo; LARGE = nf-compute-200 (2 dedicated vCPU / 4GB) $48/mo. All 24/7 always-on fixed compute plans. (Page lists the large plan as "nf-compute-200"; the input's "nf-compute-200-4" label refers to the same 2 vCPU / 4GB tier at the same $48/mo.) | $24 | $6 | $30 | 2026-07-16 |
RenderⓘSMALL = Starter instance (0.5 vCPU / 512 MB) $7/mo. MEDIUM = Standard instance (1 vCPU / 2 GB) $25/mo. LARGE = Pro instance (2 vCPU / 4 GB) $85/mo. Render bills per-second but the flat plan price is the effective 24/7 monthly cost. Larger tiers also exist (Pro Plus 4 CPU/8 GB, Pro Max 4 CPU/16 GB, Pro Ultra 8 CPU/32 GB ~$450/mo). | $25 | $11.25 | $36.25 | 2026-07-16 |
RailwayⓘRailway has no fixed instance sizes — it is fully usage-based (per-second, billed per-minute). Official rates: $0.000463/vCPU/min (~$20.28/vCPU/mo) + $0.000231/GB-RAM/min (~$10.12/GB/mo) at 730h/mo. Sizes computed at a 1:2 vCPU:GB ratio: Small = 0.5 vCPU / 1 GB = ~$20.26/mo; Medium = 1 vCPU / 2 GB = ~$40.52/mo; Large = 2 vCPU / 4 GB = ~$81.03/mo. These assume the service continuously consumes its full CPU+RAM allocation 24/7; because Railway meters ACTUAL usage, real bills are lower when CPU/RAM aren't saturated. Volumes are extra ($0.15/GB/mo). Hobby ($5/mo) and Pro ($20/mo) subscriptions each include equal usage credits, so at these usage levels the total bill ≈ the usage cost shown (subscription fee is absorbed by the credit). | $40.52 | $5 | $45.52 | 2026-07-16 |
HerokuⓘCedar Common Runtime dynos (no Private Space fee), flat monthly, one dyno running 24/7. SMALL = Standard-1X (0.5GB RAM, shared 1x-4x CPU) $25/mo — good fit for 0.5 vCPU/512MB. MEDIUM = Standard-2X (1GB RAM, shared 2x-8x CPU) $50/mo — under the 2GB target; the next Cedar tier reaching >=2GB is Performance-M (2.5GB) at $250. LARGE = Performance-M (2.5GB RAM, dedicated compute) $250/mo — undershoots the 4GB target; Performance-L (14GB) is the next step at $500. Heroku Cedar dynos bundle CPU and RAM in fixed sizes that do not map cleanly to arbitrary vCPU/RAM ratios. | $50 | — | $50 | 2026-07-16 |
No verified price yet for this size: AWS Amplify Hosting, Vercel, Netlify, Cloudflare Pages, Platform.sh, Coolify, Dokploy, Dokku, CapRover, Contabo, Supabase, Neon, PlanetScale, MongoDB Atlas, Turso, CockroachDB, Firebase, Google Cloud Run, Cloudflare R2, Backblaze B2, Wasabi, Bunny Storage, Amazon S3, DigitalOcean Spaces, Better Stack, UptimeRobot, Checkly, Pingdom, Uptime Kuma, Contabo VPS (self-host).
Serverless/edge hosts (Vercel, Netlify, Cloudflare Pages…) and managed databases are billed per request, build, seat or storage — not per always-on instance — so they show “no verified price” here rather than a misleading number. See each platform’s detail page for its own pricing model.
Egress reality check
The instance price is only half the bill — outbound bandwidth (egress) is where “cheap” hosts quietly get expensive. Here is the per-GB rate across providers, cheapest first: moving 1 TB can cost anywhere from about $1 to $150.
| Provider | Egress $/GB | Included / mo | 1 TB over the allowance |
|---|---|---|---|
| Hetzner Cloud (self-host) | $0.0012 | 20,000 GB | $1.2 |
| Akamai (Linode) | $0.005 | 2,000 GB | $5 |
| DigitalOcean Droplet (self-host) | $0.01 | 2,000 GB | $10 |
| Vultr | $0.01 | 2,000 GB | $10 |
| Fly.io | $0.02 | — | $20 |
| Koyeb | $0.02 | — | $20 |
| DigitalOcean App Platform | $0.02 | 200 GB | $20 |
| Cloudways | $0.02 | 2,000 GB | $20 |
| Railway | $0.05 | — | $50 |
| Northflank | $0.06 | — | $60 |
| Render | $0.15 | 25 GB | $150 |
Object storage flips this — Cloudflare R2, Backblaze B2 and Wasabi charge little or no egress. If bandwidth dominates your bill, switching where you serve files is usually a bigger lever than the compute price.