Cloud and AI pricing, with the arithmetic done
Every rate here is a published list price, linked to the vendor's own pricing page and stamped with the month it was checked. No affiliate links, no sponsored placements, no estimates presented as facts.
- Services priced
- 77
- AI models priced
- 25
- Comparisons
- 499
- Providers
- 54
Prices last checked . Vendors change pricing without notice — every figure links to its source so you can confirm.
Start with a question
The rankings answer the superlative queries; the comparisons answer the head-to-heads.
- Cheapest CDN providers by egress price11 entries, ranked
- Cheapest object storage per GB-month12 entries, ranked
- Cheapest VPS and cloud servers11 entries, ranked
- Cheapest cloud egress rates, all categories45 entries, ranked
- Cloud services with no egress fees6 entries, ranked
- Cheapest managed database hosting11 entries, ranked
Cheapest AI APIs (2026)
Blended cost per million tokens at three input tokens per output token.
| Model | Input /1M | Output /1M | Blended |
|---|---|---|---|
| GPT-5 nano | $0.050 | $0.400 | $0.138 |
| Mistral Small | $0.100 | $0.300 | $0.150 |
| Gemini 2.5 Flash-Lite | $0.100 | $0.400 | $0.175 |
| DeepSeek-V4-Flash | $0.140 | $0.280 | $0.175 |
| GPT-4o mini | $0.150 | $0.600 | $0.262 |
Browse cloud pricing by category
- CDN & edge deliveryContent delivery networks cache your files in points of presence near your users. Almost the entire bill is data transfer out, so the per-gigabyte egress rate — which varies by more than 100x across this category — decides what you pay.11 providers priced
- Cloud compute & VPSVirtual machines billed by the hour or by the month. Two pricing models coexist here: flat monthly plans with bundled traffic, and per-hour on-demand rates with metered egress on top. The second looks cheaper per vCPU and is usually not.15 providers priced
- GPU cloudRented accelerators for training and inference, billed by the GPU-hour. The spread between the specialist hosts and the hyperscalers is the widest of any category on this site — often 3x to 5x for the identical card — because the specialists sell capacity and the hyperscalers sell it bundled with everything else.12 providers priced
- Object storageS3-compatible buckets billed on three separate axes: gigabyte-months stored, gigabytes transferred out, and API requests. Providers compete by zeroing one axis and recovering it on another, so the cheapest option depends entirely on your read pattern.12 providers priced
- Serverless functionsCode that runs per request with no instance to provision, billed on invocations and on memory held for the duration of execution. Free tiers are wide enough that many production workloads never generate a bill at all.8 providers priced
- Application platformsManaged hosting that builds and runs your application from a git push. You pay a seat or instance fee for the convenience, and then a bandwidth rate that is typically several times what a plain CDN would charge for the same gigabytes.8 providers priced
- Managed databasesHosted Postgres, MySQL, Redis, and document stores. The category splits between provisioned instances billed per hour whether or not you use them, and serverless engines that scale to zero and bill on actual reads and writes.11 providers priced
Work out your own numbers
The calculators run on the same rate cards as every table on the site.