NVIDIA · Ada Lovelace

L40S48 GB GDDR6

A flexible accelerator for inference, rendering, and multimodal workloads.

Lowest observed rate$0.69per GPU / hour · sample feedView L40S on Flatkey Compute ↗
ArchitectureAda Lovelace
VRAM48 GB
Memory bandwidth864 GB/s
Published2023
Price board

2 configurations tracked

Median sample rate: $0.74 per GPU-hour.

#1
HyperstackGPU cloud
Configuration1× L40S · 16 vCPU
RegionCanada
PurchaseOn-demand
$0.69/ GPU / hr
#2
RunpodGPU cloud
Configuration1× L40S · 16 vCPU
RegionGlobal
PurchaseSpot
$0.79/ GPU / hr