Comparable rates by provider
Fresh, exact, full-GPU on-demand observations. Counts describe this collection’s scope.
COMPUTE ECONOMICS
The price of GPU compute.
Published rental rates, preserved in context. Compare the hardware, the bundle, and the terms behind each quote.
Uses existing Pulse authentication. The key stays in this tab’s memory.
Automatic collection is inactive. This dashboard reads saved observations; opening it does not refresh provider data.
What these prices include
A normalized rate is a compute rental cost per physical GPU. A multi-GPU node may be the minimum purchase.
Region, hardware variant, CPU/RAM bundle and service terms can differ. Unknown capacity remains unknown.
Taxes, usage-dependent storage/network, and optional software/support are excluded unless specified in the detail view.
Observed price history
Points are actual checks. Historical reconstructions are labeled separately. Close-together checks do not establish multi-day trends.
| Provider | GPU / variant | Region | Rental type / tier | GPUs / node | USD / GPU-hr ↕ | Node USD / hr | 7d / 30d | Availability | Last checked | Source |
|---|
“From” prices, quote-only offers, fractional GPUs, inference products and commitments appear only when included. Strict filtering requires known fields; it does not imply equal performance.
Evidence-backed events
Newly observed series are discovery events. Missing rows or failed requests are never reported as out of stock.
Provider-reported capacity
Stock, account-scoped reports, advisory scores and regional support are separate evidence. A current observation does not guarantee deployment.
| Provider / scope | GPU / native count | Region / tier / mode | Evidence / state | Count / bound / score | Compatible price | Source as-of / checked | Source |
|---|
Coverage, not assumptions.
Every core target and optional provider attempted. A public price is not a capacity check. Region scope varies by source.