Compare AI infrastructure for what you want to run
We may earn commissions when you shop through the links below. Learn more
GPU clouds, inference platforms, AI agent hosts and AI VPS from 10 providers. Every price links to its source and shows the date we checked it.
Not sure what AI hosting you need?
We'll help you find the right setup, even if you're completely new to AI hosting.
Or start with what you want to do:
Know what you want to compare?
Filter verified prices yourself.
Start from: H100 hostingH200 hostingL40S hostingvLLM hostingEU data residencyPer-second billing
What do you want to run?
Train or fine-tune a model
Rent GPUs by the second or the hour for training and fine-tuning runs.
Serve a model behind an API
Inference platforms and GPU servers for serving open-weight models.
Host an AI agent
Servers and managed platforms for agents that run around the clock.
Run your own model on a VPS
Monthly virtual servers for small models, agents and private AI.
Not sure which fits? Help me choose
On-demand GPU prices
Per GPU-hour, on-demand list price, excl. VAT, shown in USDEUR.
| Provider | Price per GPU-hour | vs lowest | Billing | Region | Checked | Source |
|---|---|---|---|---|---|---|
| Hyperstack | $2.50≈ €2.26 (estimated conversion from $2.50) | Lowest listed | H100 PCIe · Per minute | North AmericaN. America | Source | |
| OVHcloud | ≈ $3.19 (converted from the EUR list price, €2.80)€2.80 | +$0.69≈ €0.63 (estimated conversion from $0.69) vs lowest | H100 PCIe · Per second | Europe (EU)Europe (EU) | Source | |
| Scaleway | ≈ $3.27 (converted from the EUR list price, €2.87)€2.87 | +$0.77≈ €0.69 (estimated conversion from $0.77) vs lowest | H100 PCIe · Per minute | Europe (EU)Europe (EU) | Source | |
| Lambda | $3.29≈ €2.97 (estimated conversion from $3.29) | +$0.79≈ €0.71 (estimated conversion from $0.79) vs lowest | H100 PCIe · Per minute | Europe, North America, AsiaEurope, N. America, Asia | Source | |
| DigitalOcean | $4.41≈ €3.98 (estimated conversion from $4.41) | +$1.91≈ €1.72 (estimated conversion from $1.91) vs lowest | H100 SXM · Per second | Europe (EU), North AmericaEurope (EU), N. America | Source | |
| Nebius | $4.50≈ €4.06 (estimated conversion from $4.50) | +$2.00≈ €1.80 (estimated conversion from $2.00) vs lowest | H100 SXM · Per second | Europe (EU)Europe (EU) | Source |
| Provider | Price per GPU-hour | vs lowest | Billing | Region | Checked | Source |
|---|---|---|---|---|---|---|
| DigitalOcean | $4.47≈ €4.03 (estimated conversion from $4.47) | Lowest listed | H200 SXM · Per second | North AmericaN. America | Source | |
| Nebius | $5.40≈ €4.87 (estimated conversion from $5.40) | +$0.93≈ €0.84 (estimated conversion from $0.93) vs lowest | H200 SXM · Per second | Europe (EU), North AmericaEurope (EU), N. America | Source |
| Provider | Price per GPU-hour | vs lowest | Billing | Region | Checked | Source |
|---|---|---|---|---|---|---|
| Nebius | $1.55≈ €1.40 (estimated conversion from $1.55) | Lowest listed | L40S · Per second | Europe (EU)Europe (EU) | Source | |
| DigitalOcean | $1.57≈ €1.42 (estimated conversion from $1.57) | +$0.02≈ €0.02 (estimated conversion from $0.02) vs lowest | L40S · Per second | North AmericaN. America | Source | |
| OVHcloud | ≈ $1.60 (converted from the EUR list price, €1.40)€1.40 | +$0.05≈ €0.04 (estimated conversion from $0.05) vs lowest | L40S · Per second | Europe (EU)Europe (EU) | Source | |
| Scaleway | ≈ $1.68 (converted from the EUR list price, €1.47)€1.47 | +$0.13≈ €0.12 (estimated conversion from $0.13) vs lowest | L40S · Per minute | Europe (EU)Europe (EU) | Source |
| Provider | Price per GPU-hour | vs lowest | Billing | Region | Checked | Source |
|---|---|---|---|---|---|---|
| Lambda | $6.99≈ €6.31 (estimated conversion from $6.99) | Lowest listed | B200 SXM · Per minute | Europe, North America, AsiaEurope, N. America, Asia | Source | |
| Nebius | $8.50≈ €7.67 (estimated conversion from $8.50) | +$1.51≈ €1.36 (estimated conversion from $1.51) vs lowest | B200 SXM · Per second | North America, AsiaN. America, Asia | Source | |
| Together AI | $8.99≈ €8.11 (estimated conversion from $8.99) | +$2.00≈ €1.80 (estimated conversion from $2.00) vs lowest | B200 SXM · Per minute | Europe, North AmericaEurope, N. America | Source |
| Provider | Price per GPU-hour | vs lowest | Billing | Region | Checked | Source |
|---|---|---|---|---|---|---|
| Hyperstack | $1.35≈ €1.22 (estimated conversion from $1.35) | Lowest listed | A100 PCIe 80GB · Per minute | North AmericaN. America | Source | |
| Lambda | $1.99≈ €1.80 (estimated conversion from $1.99) | +$0.64≈ €0.58 (estimated conversion from $0.64) vs lowest | A100 40GB · Per minute | Europe, North America, AsiaEurope, N. America, Asia | Source |
| Provider | Price per GPU-hour | vs lowest | Billing | Region | Checked | Source |
|---|---|---|---|---|---|---|
| OVHcloud | ≈ $0.855 (converted from the EUR list price, €0.75)€0.75 | Lowest listed | L4 · Per second | Europe (EU)Europe (EU) | Source | |
| Scaleway | ≈ $0.898 (converted from the EUR list price, €0.788)€0.788 | +$0.04≈ €0.04 (estimated conversion from $0.04) vs lowest | L4 · Per minute | Europe (EU)Europe (EU) | Source |
| Provider | Price per GPU-hour | vs lowest | Billing | Region | Checked | Source |
|---|---|---|---|---|---|---|
| OVHcloud | ≈ $0.867 (converted from the EUR list price, €0.76)€0.76 | Lowest listed | A10 · Per second | Europe (EU)Europe (EU) | Source | |
| Lambda | $1.29≈ €1.16 (estimated conversion from $1.29) | +$0.42≈ €0.38 (estimated conversion from $0.42) vs lowest | A10 · Per minute | Europe, North America, AsiaEurope, N. America, Asia | Source |
| Provider | Price per GPU-hour | vs lowest | Billing | Region | Checked | Source |
|---|---|---|---|---|---|---|
| Hyperstack | $0.50≈ €0.451 (estimated conversion from $0.50) | Lowest listed | RTX A6000 · Per minute | North AmericaN. America | Source | |
| Lambda | $1.09≈ €0.984 (estimated conversion from $1.09) | +$0.59≈ €0.53 (estimated conversion from $0.59) vs lowest | RTX A6000 · Per minute | Europe, North America, AsiaEurope, N. America, Asia | Source |
How this comparison works
Prices come from official pages.
441 sources, each linked next to the value it supports.
MethodologyEvery price shows when we checked it.
Latest check 20 Oct 2026. 10 of 296 published price fields are checked automatically; the rest in research passes.
Price statusSome provider links are affiliate links.
1 provider has an active affiliate link. If you sign up or buy through it, we may earn a commission; that never changes the order of results or the data we show.
Affiliate disclosureOrder is never for sale.
Lists sort by your filters, price or name.
How we order
How does AIHostCompare make money? Read the affiliate disclosure.
Tools
GPU cost calculator
Estimate a month of GPU time from 114 verified prices.
Example: a month of H100 80GB time at the 3 lowest list prices
Estimated monthly compute for 1× H100 80GB, 8 h/day × 22 days (176 GPU-hours) at each provider's lowest single-GPU on-demand list price. The 3 lowest of 7 providers (range $440≈ €397 (estimated conversion from $440)–$966≈ €872 (estimated conversion from $966)). Compute only. Provider Per month Checked Source Hyperstack pricing $440≈ €397 (estimated conversion from $440) H100 PCIe 80GB · Source OVHcloud pricing ≈ $562 (converted from the EUR list price, €493)€493 H100 PCIe 80GB · Source Scaleway pricing ≈ $575 (converted from the EUR list price, €505)€505 H100 PCIe 80GB · Source VRAM and model fit
Check which GPU a model needs · 11 open-weight models.
Agent hosting selector
Answer a few questions about your agent and get a hosting category and a shortlist.
Compare providers
Pick up to 4 providers and see prices, regions and features side by side.
Browse by category, GPU and framework
Categories
GPUs
- H100 hosting
- H200 hosting
- L40S hosting
- RTX 4090 cloud
Still deciding?
Answer a few questions for a shortlist with sources, or browse all 10 providers.