Plan your infrastructure on evidence, not marketing.
Hosting, databases and the infrastructure between them, read from what providers publish — dated, sourced, and never paid for. Plan what your next project needs, or check what you already have.
Find infrastructure for what you're building.
Pick everything your project is made of and needs. The list below narrows to providers that document support for everything picked — nothing here is a guess at what a provider can do.
Picks inetGeek's usual starting checklist for that stage below — every box stays yours to change.
Frameworks
Backend
Self-hosted apps
Runtimes
CMS
| Provider | Category | Documented fit | Entry paid plan | Free tier |
|---|---|---|---|---|
| AWS Amplify | Frontend & serverless | No monthly minimum — Amplify Hosting is billed per unit. A representative month for one small app, priced at Amplify's published rates: 100 build minutes ($1.00), 5 GB stored ($0.11), 100 GB served ($15.00), with SSR requests inside the free 500,000 — about $16.11. Our arithmetic on Amplify's rates, not a price Amplify quotes, and it moves with traffic: egress is 93% of it. Recorded without a magnitude for that reason. | Yes | |
| Cloudflare Pages | Frontend & serverless | $5 / month account minimum | Yes | |
| DigitalOcean | VPS & cloud | $4.00 | Not documented | |
| DreamHost | Mass-market | $10.99/month on renewal for Web Hosting Launch, the cheapest shared plan. The advertised $2.89 is a first-year rate on an annual prepay. | Not documented | |
| Fly.io | App platforms | Not documented | Not documented | |
| Flywheel | Managed hosting | $25/month for Starter, billed $300 a year with two months free. Monthly billing is offered at a higher rate. | Not documented | |
| Google Cloud Run | App platforms | Not documented | Yes | |
| Heroku | App platforms | $5 per month (Eco dynos plan, 1000 dyno hours shared across the account) | Not documented | |
| Hetzner | VPS & cloud | €5.99/month for the CX23 (2 vCPU, 4 GB, 40 GB NVMe), including a primary IPv4 address, or €0.0096/hour. The ARM CAX11 with identical specs is €6.49. Converted for comparison at the ECB reference rate of 1 EUR = 1.1622 USD (2026-09-04, api.frankfurter.dev), which puts the CX23 at about $6.96/month. Hetzner publishes no USD price; the conversion is ours and moves with the rate. | Not documented | |
| Hostinger | Mass-market | $10.99/month on renewal (Premium plan) | Not documented | |
| Hostinger VPS | VPS & cloud | $11.99/month on renewal for KVM 1, on a two-year term. The advertised $6.49 is an introductory rate for the first term only. | Not documented | |
| Kinsta | Managed hosting | $35/month for the entry single-site plan (first month $0, then $35/month); $350/year if paid annually | Not documented | |
| Linode (Akamai Cloud) | VPS & cloud | $5.00/month (Nanode 1 GB shared CPU plan, North America pricing) | Not documented | |
| Netlify | Frontend & serverless | $9/month for 1,000 credits/month | Yes | |
| Pressable | Managed hosting | $20.83/month for Signature 1, billed $250 a year. Monthly billing is offered at a higher rate; the yearly plan advertises two months free. | Not documented | |
| Railway | App platforms | $5 / month (Hobby) | Yes | |
| Render | App platforms | Not documented | Yes | |
| SiteGround | Mass-market | $17.99/month on renewal ($3.99/mo intro, prepaid 12 months) | Not documented | |
| Vercel | Frontend & serverless | $20 per developer seat / month (Pro) | Yes | |
| Vultr | VPS & cloud | $2.50/mo (Regular Performance Cloud Compute, IPv6-only plan) | Not documented | |
| WP Engine | Managed hosting | Starting at $28.00/month for the Managed Hosting Platform, quoted in USD. WP Engine's own footnote says first-year pricing applies to new Essential customers and that renewal pricing is subject to change; no renewal figure is published on the plans page. | Not documented |
Places to run your project, not what it stores its data in or how it signs a user in — every other layer your project might need is below, collapsed until you need it.
Storing what your project produces?
Object storage is a separate decision from where your app runs, and egress is the sharpest line between the providers documented here — free to serve back out for most of them, and a real per-GB charge from the hyperscalers in the set, which is the one column below actually worth comparing.See the full list →
| Provider | Category | Documented fit | Egress over allowance | Free tier |
|---|---|---|---|---|
| Amazon S3 | Object storage | $0.09 per GB for the first 10 TB a month out to the internet, after a free 100 GB aggregated globally. It falls to $0.085 above 10 TB, $0.07 above 50 TB and $0.05 above 150 TB. This is AWS's outbound data transfer rate rather than an S3-specific one — the charge that R2, Wasabi and Tigris do not levy at all. | Yes | |
| Azure Blob Storage | Object storage | First 100 GB a month free, then $0.087 per GB over Microsoft's global network, falling to $0.083 above ~10 TB, $0.07 above ~50 TB and $0.05 above ~150 TB. Routing over the public internet instead is $0.08 at the first tier. This is Azure-wide bandwidth rather than a Blob-specific charge. | Not documented | |
| Backblaze B2 | Object storage | Free up to three times your stored capacity each month on the $6.95/TB tier; unlimited free egress on the $15/TB tier. Egress beyond the 3x allowance is chargeable. | Yes | |
| Bunny Storage | Object storage | Free to Bunny's own CDN, and no API fees or API egress charges. Bunny prices storage by replication rather than by transfer. | Not documented | |
| Cloudflare R2 | Object storage | Free. Cloudflare charges nothing for egress to the internet on any storage class; Infrequent Access adds a $0.01/GB retrieval fee instead. | Yes | |
| IDrive e2 | Object storage | Free up to three times your active storage volume each month — the same shape as Backblaze B2's allowance rather than the unconditional free egress R2 and Wasabi advertise. | Not documented | |
| Storj | Object storage | $7 per TB, the same rate as storage — reduced from $20 per TB on the Standard tier. Storj charges for egress where R2, Wasabi and Tigris do not, but at a tenth of what S3 charges. | Not documented | |
| Tigris | Object storage | None. Tigris charges no egress fees; retrieval is free on Standard and Archive tiers and $0.01–$0.03 per GB on the Infrequent Access and Glacier tiers. | Yes | |
| Wasabi | Object storage | None. Wasabi charges for capacity only, with no egress or API request fees. | No |
Need somewhere to put your data?
Managed databases split on a couple of real questions: whether you can branch a database the way you branch a git repo, and whether read replicas exist for scaling reads separately from writes.See the full list →
| Provider | Category | Documented fit | Entry paid plan | Free tier |
|---|---|---|---|---|
| Aiven | Managed database | $5 a month for Developer — one VM, 1 CPU, 1 GB RAM, 8 GB storage, with no choice of cloud or region and no connection pooling. The first plan that behaves like a production service is Startup at $75 a month or $0.10 an hour, which is where the uptime SLA and multi-CPU sizing begin; Hobbyist sits between them at $12. | Yes | |
| Amazon Aurora | Managed database | Not documented | Yes | |
| Amazon RDS | Managed database | Not documented | Not documented | |
| Crunchy Bridge | Managed database | $9 a month for Hobby-0 — 2 cores, 512 MB, AWS only. Crunchy says the Hobby tier runs on lower-quality infrastructure, has no PgBouncer and is not intended for production; the first production tier is Standard-4 at $70 a month for 1 core and 4 GB. There is no free plan. | Not documented | |
| Neon | Managed database | No fixed monthly fee. Launch is usage-based — Neon's own worked example puts typical spend at $15/month for intermittent load on 1 GB — so there is no entry price to compare, only a rate card. Recorded without a magnitude because a vendor's spend estimate is not a price. | Yes | |
| Nile | Managed database | $15/month for Pro, which carries a production SLA. Scale is $350/month. | Not documented | |
| PlanetScale | Managed database | $5/month for a single-node Postgres database on network-attached storage; PlanetScale Metal with local NVMe starts at $50/month | No | |
| Prisma Postgres | Managed database | $10/month for Starter. Pro is $49 and Business $129, each including a larger request allowance. | Yes | |
| Supabase | Managed database | $25/month for the Pro plan, including the first project and $10/mo of compute credits; additional projects from $10/mo | Yes | |
| TigerData | Managed database | Compute starts at $30/month on Performance, billed hourly, with storage charged separately. Scale starts at $36/month. | Not documented | |
| Turso | Managed database | $4.99/month for Developer, the cheapest paid plan. Scaler is $24.92 and Pro $416.58. | Yes |
Watching for what breaks in production?
Error tracking splits on two real questions: whether session replay exists alongside error capture, and whether you can self-host it instead of sending your data to someone else.See the full list →
| Provider | Category | Documented fit | Entry paid plan | Free tier |
|---|---|---|---|---|
| Axiom | Observability | $25/month platform fee for Cloud, plus usage, with no minimum commitment. The plan carries an always-free allowance of 1 TB loaded, 100 GB-hours of query compute and 100 GB stored. | Yes | |
| Better Stack | Observability | From $29/month, or $34/month depending on the plan selected; Better Stack advertises both figures as starting prices on the same page. | Yes | |
| Datadog | Observability | $15 per host per month for Infrastructure Pro, the entry paid product. Datadog prices per product rather than per plan, so this is one meter of many rather than a platform fee. | Yes | |
| GlitchTip | Observability | $15/month for the 100,000 event plan; $50 for 500,000 and $250 for 3 million | Yes | |
| Grafana Cloud | Observability | $19/month platform fee for Pro, which includes 10,000 active series before usage billing starts. | Yes | |
| OpenObserve | Observability | Not documented | Not documented | |
| Sentry | Observability | $26/month for Team, the cheapest paid plan | Yes | |
| SigNoz | Observability | $49/month for Teams, and the $49 is usage credit rather than a fee on top — SigNoz bills against it before charging more. | No |
Calling a model from your app?
LLM APIs are priced per token, not per month, so the entry price below is the cheapest model's rate rather than a plan fee. What actually varies between providers is whether that price includes prompt caching.See the full list →
| Provider | Category | Documented fit | Cheapest model, per token | Free tier |
|---|---|---|---|---|
| Anthropic | LLM API | $1 per million input tokens for Claude Haiku 4.5, the cheapest current model, with output at $5. | Not documented | |
| Baseten | LLM API | $0.10 per million input tokens for GPT OSS 120B, with output at $0.50. GLM-5.3-Flash is $0.15 in and $0.50 out. | Not documented | |
| DeepInfra | LLM API | $0.06 per million input tokens for DeepSeek-V4-Flash-0731, with cached input at $0.015 and output at $0.18. DeepSeek itself charges $0.44 at peak for the same checkpoint and Fireworks $0.22. | Not documented | |
| DeepSeek | LLM API | $0.44 per million input tokens for deepseek-v4-flash at peak on a cache miss, $0.22 off-peak, with output at $1.32 peak. | Not documented | |
| Fireworks AI | LLM API | $0.15 per million input tokens for GLM 5.3 Flash, with output at $0.50. DeepSeek-V4-Flash is $0.22 in and $0.66 out — the same model DeepSeek itself sells at $0.44 peak. | Yes | |
| Google Gemini API | LLM API | $0.30 per million input tokens for Gemini 3.5 Flash-Lite, with output at $2.50. | Yes | |
| Novita AI | LLM API | $0.075 per million input tokens for GLM 5.3 Flash at short context, rising to $0.15 above the threshold, with cache reads at $0.015 and output at $0.25. Fireworks serves the same model at $0.15. | Not documented | |
| OpenAI | LLM API | $0.20 per million input tokens for gpt-5.6-luna at short context, with output at $1.20. | Not documented |
Sending transactional email?
Transactional email splits on deliverability control — a dedicated sending IP — and whether the provider can receive mail back for your app, not just send it.See the full list →
| Provider | Category | Documented fit | Entry paid plan | Free tier |
|---|---|---|---|---|
| AhaSend | Transactional email | $10/month for Pro; Max is $80/month. | Yes | |
| Amazon SES | Transactional email | No monthly plan — SES bills per email. Ten thousand emails, the tier Postmark and Mailgun both price at, costs $1.00 at the published $0.10 per 1,000. Our arithmetic on Amazon's rate, not a price Amazon quotes, and attachments add $0.12 per GB. | Yes | |
| MailerSend | Transactional email | $31.50/month for Starter, billed $378 a year. Professional is $99/month, billed $1,188. | Not documented | |
| Mailgun | Transactional email | $15/month for Basic, described as a starting price. | Yes | |
| Postmark | Transactional email | $15.00/month for Basic at the 10,000-email tier. Postmark prices by volume, so the figure rises with the slider. | Yes | |
| Resend | Transactional email | $20/month for Pro, the cheapest paid plan. | Yes | |
| SMTP2GO | Transactional email | $10/month for Starter, or $100 a year. Professional is $75/month. | Yes |
Handling sign-in and user accounts?
Auth is where the SSO tax hides: SAML support, SCIM provisioning and MFA are the three capabilities most often paywalled well above the entry price shown below.See the full list →
| Provider | Category | Documented fit | Entry paid plan | Free tier |
|---|---|---|---|---|
| Auth0 | Auth | $35/month for Essentials at 500 monthly active users; Professional is $240/month at the same volume. Both scale with the MAU slider. | Yes | |
| Better Auth | Auth | $20/month for Pro on the managed infrastructure. The framework itself costs nothing — the plan buys a dashboard, audit logs and security detections around it. | Yes | |
| Clerk | Auth | $25/month for Pro, or $20/month billed annually. Business is $300/month. | Yes | |
| Kinde | Auth | $25/month for Pro. Plus is $75 and the top tier $250 — each including the same 10,500 MAU as the free plan. | Yes | |
| Stytch | Auth | Not documented | Yes | |
| SuperTokens | Auth | $100/month minimum billing on the managed cloud, which covers the first 4,999 MAU at $0.02 each. | Not documented | |
| WorkOS | Auth | Not documented | Yes |
Need background jobs or a queue?
Queues and background job runners split on whether a failed message lands in a dead-letter queue you can inspect, or just vanishes.See the full list →
| Provider | Category | Documented fit | Entry paid plan | Free tier |
|---|---|---|---|---|
| Amazon SQS | Queue | Not documented | Yes | |
| Azure Queue Storage | Queue | Not documented | Not documented | |
| Cloudflare Queues | Queue | Not documented | Yes | |
| Google Cloud Pub/Sub | Queue | Not documented | Yes | |
| Hatchet | Queue | $500/month for Team, which adds users, tenants, retention and throughput on top of usage. Scale is $1,000/month. | Yes | |
| Inngest | Queue | $99/month for Pro, described as a starting price with usage above the included allowance billed on top. | Yes | |
| Temporal | Queue | $100/month for Essentials, including 1 million Actions, 1 GB active storage and 40 GB retained storage. Business starts at $500/month with 2.5 million Actions. | Yes | |
| Trigger.dev | Queue | $10/month for Hobby, which includes $10 of credits; Pro is $50/month including $50 of credits. | Yes | |
| Upstash QStash | Queue | Not documented | Yes | |
| Zeplo | Queue | $39/month for Company, including 1,000,000 requests and three team members. Additional members are $15 each. | Yes |
Training or serving a model that needs a GPU?
GPU compute is priced per hour, per card, and the rate for the same H100 SXM varies sharply across the providers below — a marketplace like Vast.ai prices well under a dedicated cloud — no capability chips here, since none of these differ on what they support, only on what they charge.See the full list →
| Provider | Category | Documented fit | H100 SXM, per GPU-hour | Free tier |
|---|---|---|---|---|
| Baseten | LLM API | $6.50 per hour for a dedicated H100 80GB. Baseten publishes $0.10833 per MINUTE — the hourly figure is that times 60, ours rather than Baseten's, and the page offers an hourly toggle. A100 80GB is $0.06667 a minute and B200 $0.16633. | Not documented | |
| CoreWeave | GPU compute | $6.16 per GPU-hour for H100 SXM 80GB. CoreWeave sells the 8-GPU HGX H100 node at $49.24/hour on demand — the per-GPU figure is that divided by eight, ours rather than CoreWeave's, and you cannot rent one card. | Not documented | |
| DeepInfra | LLM API | $2.20 per GPU-hour for a dedicated H100 80GB — below every dedicated cloud in this dataset. A100 80GB is $0.89, H200 $2.69, B200 $3.69. Billed in minute granularity, invoiced weekly. | Not documented | |
| Fireworks AI | LLM API | $7.00 per hour for an H100 80GB on demand, or $8.00 on the shorter commitment shown beside it. The same rate applies to an H200 141GB. | Yes | |
| Hyperstack | GPU compute | $3.20 per hour for H100 SXM 80GB on demand. The NVLink variant is $2.60 and the PCIe H100 is $2.50. | Not documented | |
| Lambda | GPU compute | $4.29 per GPU-hour for H100 SXM 80GB on a single-GPU instance, falling to $3.99 on an 8-GPU node. The PCIe variant is $3.29. | Not documented | |
| Novita AI | LLM API | $3.39 per GPU-hour for an on-demand H100 SXM 80GB, or $1.70 spot. The page names the card the site's GPU comparison is fixed on, so no substitution is involved. | Not documented | |
| RunPod | GPU compute | $3.49 per hour for H100 SXM 80GB, single card. The PCIe variant is $2.89 and H100 NVL is $3.19. | Not documented | |
| Vast.ai | GPU compute | Median $2.16 per hour for H100 SXM 80GB, with a floor of $1.33 from the cheapest host. Vast is a marketplace — the floor is one host's offer at one moment and the median is what the market is actually charging, so both move without anyone announcing anything. This figure is a snapshot, not a rate card. | Not documented |
Compare infrastructure providers.
Objective comparisons across pricing, infrastructure and developer experience — built from published documentation rather than marketing copy.
| DigitalOcean | Hetzner | |
|---|---|---|
| Entry paid plan | $4.00 | €5.99/month for the CX23 (2 vCPU, 4 GB, 40 GB NVMe), including a primary IPv4 address, or €0.0096/hour. The ARM CAX11 with identical specs is €6.49. Converted for comparison at the ECB reference rate of 1 EUR = 1.1622 USD (2026-09-04, api.frankfurter.dev), which puts the CX23 at about $6.96/month. Hetzner publishes no USD price; the conversion is ours and moves with the rate. |
| Pricing model | billed per second with a minimum charge of 60 seconds or $0.01, whichever is higher | Your server's bill will never exceed its monthly price cap. If you delete your cloud server before the end of the billing month, we will only bill you for the hourly rate. |
| Egress over allowance | $0.01 per GiB (Droplets); $0.02 per GiB (App Platform) | €1.00 per TB over the included traffic in Europe and the US; €7.40 per TB in Singapore. Billed in blocks of 100 MB, rounded up. |
| When you exceed a cap | Billed automatically — outbound transfer beyond the pooled plan allowance is charged at $0.01 per GiB; inbound transfer is free | Billed automatically — traffic beyond the amount included with the server is billed in blocks of 100MB, rounded up; Hetzner emails at 75% and 100% of the included traffic but explicitly does not cap or stop usage |
| Storage (entry plan) | 10 GiB SSD on the cheapest Basic Droplet (Regular CPU, 512 MiB / 1 vCPU, $4.00/mo, size slug s-1vcpu-512mb-10gb) | 40 GB NVMe on CX23 and CAX11, the smallest plans in Hetzner's cost-optimized line |
| Memory (entry plan) | 512 MiB on the cheapest Basic Droplet (Regular CPU, 1 vCPU, $4.00/mo, size slug s-1vcpu-512mb-10gb) | 4 GB on CX23 and CAX11, the smallest plans in Hetzner's cost-optimized line |
Infrastructure decisions without marketing copy.
Every comparison is assembled from the same five dimensions, so two pages can be read against each other.
- Pricing
- Published plan pricing and the infrastructure costs underneath it, read from official pricing pages.
- Limits
- Bandwidth, compute, storage, build minutes and deployment ceilings — including the ones that only appear in the docs.
- Developer experience
- Deployments, logs, CLI, Git integration and observability, judged by what the workflow actually feels like.
- Infrastructure
- Regions, edge network, runtimes and scaling behaviour, rather than marketing claims about speed.
- Lock-in
- How much of the setup is portable, and what has to be rewritten to leave.
Every figure keeps the sentence it was read from. Flip between the two and the reduction is visible — not something to take on trust.
DigitalOcean, two ways
- When you exceed a cap
“Each Droplet plan includes an amount of free outbound data transfer. Additional outbound transfer is billed at $0.01 per GiB. Inbound transfer to Droplets is free.”
Billed automatically — outbound transfer beyond the pooled plan allowance is charged at $0.01 per GiB; inbound transfer is free
Droplet Pricing | DigitalOcean Documentation · read 2026-09-05
- Entry paid plan
“512 MiB / 1 vCPU / 500 GiB transfer / 10 GiB SSD — $0.00595 /hr, $4.00 /mo”
$4.00
Droplet Pricing | DigitalOcean · read 2026-09-05
- Egress over allowance
“Droplets: Additional outbound transfer is billed at $0.01 per GiB. App Platform: Additional outbound transfer is billed at $0.02 per GiB.”
$0.01 per GiB (Droplets); $0.02 per GiB (App Platform)
Bandwidth Billing | DigitalOcean Documentation · read 2026-09-05
Check the infrastructure behind any domain.
The lookup at the top of this page reads live authoritative DNS. The full tools cover every record type worth inspecting.
- A / AAAA
- CNAME
- MX
- TXT
- NS
- SOA
- CAA
- DNSSEC
- SPF
- DMARC
www.example.com · A
Popular comparisons.
Free DNS & email tools.
Free technical utilities for checking domains, DNS records and email configuration. No account required.
DNS LookupLook up A, AAAA, CNAME, MX, TXT and nameserver records for any domain.MX LookupFind the mail servers handling email for a domain.SPF CheckerInspect and validate a domain's SPF policy.DMARC CheckerFind and validate DMARC configuration.DNS PropagationCheck how a record has propagated across public resolvers.