All GPUs
DatacenterNone available
L40S
NVIDIA L40S
Ada Lovelace datacenter GPU optimized for inference.
Specifications
VRAM48 GB
Memory TypeGDDR6
Bandwidth864 GB/s
CUDA Cores18,176
Tensor Cores568
FP32 Performance91.6 TFLOPS
FP16 Performance183.2 TFLOPS
TDP350 W
ArchitectureAda Lovelace
GPU ChipAD102
Best For
LLM InferenceVideo ProcessingGenerative AI
Price per hour
$0.68
Cost Estimate
per hour$0.68
per day$16.32
per week$114.24
per month$489.60