NVIDIA A100 80GB
Datacenter GPU commonly rented for training and large-model inference. Memory figure is from NVIDIA's public product materials; cloud hourly rates come from provider pricing snapshots, not from NVIDIA.
VRAM
80 GB
TDP
400 W
Architecture
Ampere
No throughput (tokens/s) is listed — we do not hard-code unsourced performance claims.
Cloud offers
- RunPod · $1.64/hr · global · Rent on RunPod
- Vultr · $2.00/hr · ewr · Rent on Vultr