On-demand from $0.80/hr · spot from $0.59/hr · providers: aws · runpod · vast
| Provider | Region / tier | Type | $/hr | History | Provenance | |
|---|---|---|---|---|---|---|
| vast | global | spot p25 bid ×16 | $0.59/hr | source | rent | |
| vast | global | on_demand p25 ×16 | $0.80/hr | source | rent | |
| runpod | secure-cloud | on_demand list · Low stock | $1.09/hr | source | rent | |
| runpod | secure-cloud | spot list · Low stock | $1.09/hr | source | rent | |
| aws | us-east-1 | on_demand instance-bundled | $1.86/hr | source | rent |
History JSON for this GPU → · Will my model fit?
As of the latest snapshot, L40S 48GB on-demand rental starts at $0.80/hr (vast, global) and ranges up to $1.86/hr across 3 tracked providers. Prices refresh continually.
vast currently lists the lowest on-demand L40S 48GB price we track, at $0.80/hr. Every number links to the page it was observed on.
Yes — spot/interruptible L40S 48GB capacity starts at $0.59/hr (vast), versus $0.80/hr on-demand. Spot can be reclaimed by the provider; use it for fault-tolerant work.
48 GB. At fp16 that fits roughly a 20B-parameter model on a single GPU; use the fit calculator for other precisions.
PRO 6000 MIG 48GB · RTX 4090 48GB (modified) · RTX A6000 48GB · RTX PRO 5000 48GB · A100 PCIE 40GB · A100 SXM4 40GB