Hyperstack
Single-GPU to 8-GPU VMs on A100 80GB PCIe, H100 PCIe and L40 for fine-tuning and inference, billed per minute with no data-transfer fees
A6000 48 GB · from:
$0.50≈ €0.447 (estimated conversion from $0.50)
Compare GPU cloud providers
We may earn commissions when you shop through the links below. Learn more
Filter GPU offers by model, VRAM, region, billing and storage, with a check date on every price.
GPU cloud providers
Price per GPU model
The lowest verified pay-as-you-go price per GPU model among the providers on this page, with the median of the listed prices. Per GPU-hour, excluding VAT, storage and egress; a price rented as a whole node is marked. 100 GPU-hours is the hourly price times 100, not a monthly estimate; the GPU cost calculator adds storage, egress and your own hours. Models priced by one provider only are left out; prices as checked up to .
| GPU model | Lowest pay-as-you-go price | Median listed price | 100 GPU-hours at the lowest price | Compare |
|---|---|---|---|---|
| B200 | $6.00≈ €5.36 (estimated conversion from $6.00) per GPU-hour | $6.89≈ €6.16 (estimated conversion from $6.89) per GPU-hour | $600.00≈ €536.05 (estimated conversion from $600.00) | Filter the list above |
| H200 | $3.99≈ €3.56 (estimated conversion from $3.99) per GPU-hour | $4.51≈ €4.02 (estimated conversion from $4.51) per GPU-hour | $399.00≈ €356.47 (estimated conversion from $399.00) | H200 hosting prices |
| A100 SXM | $1.60≈ €1.43 (estimated conversion from $1.60) per GPU-hour | $2.50≈ €2.23 (estimated conversion from $2.50) per GPU-hour | $160.00≈ €142.95 (estimated conversion from $160.00) | Filter the list above |
| H100 PCIe | $2.50≈ €2.23 (estimated conversion from $2.50) per GPU-hour | $3.23≈ €2.89 (estimated conversion from $3.23) per GPU-hour | $250.00≈ €223.35 (estimated conversion from $250.00) | H100 hosting prices |
| H100 SXM | $3.20≈ €2.86 (estimated conversion from $3.20) per GPU-hour | $3.99≈ €3.56 (estimated conversion from $3.99) per GPU-hour | $320.00≈ €285.89 (estimated conversion from $320.00) | H100 hosting prices |
| A6000 | $0.50≈ €0.447 (estimated conversion from $0.50) per GPU-hour | $1.09≈ €0.974 (estimated conversion from $1.09) per GPU-hour | $50.00≈ €44.67 (estimated conversion from $50.00) | Filter the list above |
| L40S | $1.55≈ €1.38 (estimated conversion from $1.55) per GPU-hour | $1.72≈ €1.54 (estimated conversion from $1.72) per GPU-hour | $154.84≈ €138.34 (estimated conversion from $154.84) | L40S hosting prices |
| A100 | $1.99≈ €1.78 (estimated conversion from $1.99) per GPU-hour | $1.99≈ €1.78 (estimated conversion from $1.99) per GPU-hour | $199.00≈ €177.79 (estimated conversion from $199.00) | Filter the list above |
| A10 | ≈ $0.867 (converted from the EUR list price, €0.76)€0.76 per GPU-hour | $1.06≈ €0.95 (estimated conversion from $1.06) per GPU-hour | $86.66≈ €77.42 (estimated conversion from $86.66) | Filter the list above |
| L4 | $0.80≈ €0.715 (estimated conversion from $0.80) per GPU-hour | $0.877≈ €0.783 (estimated conversion from $0.877) per GPU-hour | $80.00≈ €71.47 (estimated conversion from $80.00) | Filter the list above |
| V100 | $0.79≈ €0.706 (estimated conversion from $0.79) per GPU-hour | $0.794≈ €0.709 (estimated conversion from $0.794) per GPU-hour | $79.00≈ €70.58 (estimated conversion from $79.00) | Filter the list above |
What the price types mean
- On-demand
No commitment. The instance is billed per second, minute or hour while it runs.
- Secure and community cloud
Some providers sell the same GPU in data-centre hosting and on third-party hosts. The community tier is usually cheaper and comes with weaker guarantees; check the terms on the provider page.
- Spot
Interruptible capacity at a lower price. The provider can reclaim it, so long jobs need checkpoints.
- Marketplace minimum and median
On marketplaces, hosts set their own prices. The minimum is the lowest listing and the median the middle listing when checked; both can change within the hour.
- Serverless
Billed only while a request or job runs, usually per second. Cold starts add latency when a worker scales up from zero.
Tools for this decision
- GPU Cost CalculatorEstimate monthly GPU spend from verified hourly prices, usage hours, instances and storage.
- VRAM / Model FitEstimate the GPU memory a model needs for inference or fine-tuning, and which GPUs fit.
- Agent Hosting SelectorAnswer a few questions about your agent and get a best-fit hosting category and provider shortlist.
How we compare
Prices per GPU-hour.
GPU prices are USD per single GPU per hour from the provider's own pricing page or API. Multi-GPU nodes are divided by their GPU count, and a price only rented as a node says so (“rented as 8 GPUs”). Rows lead with the cheapest pay-as-you-go price (on-demand, secure, community or serverless). A spot, reserved or marketplace price leads only when the provider has no pay-as-you-go price here, or when you filter by that billing; it is labelled with its price type and listed in its own group after pay-as-you-go prices, never ranked against them. Prices exclude VAT and sales tax.Not verified means not verified.
When no official source confirms a value we show “Not verified” instead of an estimate. Where a provider's terms do not allow its prices to be republished, we link to its live pricing instead.Freshness on every row.
Updated today: checked on the current day (UTC). Fresh: up to 7 days. Stale: up to 30 days. Needs refresh: older than that.No paid placement.
Results only change order with the sort you pick. Affiliate relationships never affect order, filters or which providers appear.
Filters match a provider when one of its listed offers meets every GPU, VRAM, region, price and billing-increment condition together. Spot, reserved and serverless billing list each provider that offers that pricing and show only prices on that basis, or “not verified” when we have none. Read the full methodology and our affiliate disclosure.
Frequently asked questions
Are prices per GPU or per instance?
Per GPU. Every GPU price is shown in USD per single GPU per hour; offers for multi-GPU nodes are divided by the number of GPUs, and a price you can only rent as a whole node says so (for example “rented as 8 GPUs”). Storage, data transfer and taxes are not included unless the provider includes them.
Which cloud GPU provider is the cheapest?
It depends on the GPU. On this page the lowest verified pay-as-you-go price is $3.20 per GPU-hour for the H100 SXM (Hyperstack, on-demand, rented as 8 GPUs), $3.99 per GPU-hour for the H200 (Hyperstack, on-demand, rented as 8 GPUs) and $1.55 per GPU-hour for the L40S (Nebius, on-demand), checked 20 Oct 2026. Marketplace minimums and spot prices can be lower but are not guaranteed capacity, so compare the same GPU, price type and billing increment. Amounts are in US dollars (USD), excluding VAT.
How much does a cloud GPU cost?
It depends on the GPU: verified pay-as-you-go prices on this page run from $0.50 per GPU-hour (A6000 at Hyperstack) to $8.99 (B200 at Together AI), checked 20 Oct 2026. For the H100 SXM, the most widely listed GPU here, they start at $3.20 (Hyperstack, on-demand, rented as 8 GPUs) with a median of $3.99 across 15 prices. Amounts are in US dollars (USD), excluding VAT.
Which cloud GPU is best for AI?
The one whose memory fits your model and whose price fits your utilisation. VRAM decides which models fit; compute and memory bandwidth decide speed. The VRAM / Model Fit tool shows which GPUs fit a given model before you compare prices.
Why do some values say “Not verified”?
We only show a value when an official provider source confirms it. If we could not confirm a price or feature, the row says “Not verified” instead of showing an estimate, and the provider page lists the sources we did check.