A100 Specifications
The A100 is ideal for large-scale model training and inference, providing 80-90% of the capabilities of newer H100s for many workloads at a lower cost.
80 GB
HBM2e Memory
Run 70B+ parameter models on a single GPU.
600 GB/s
NVLink Bandwidth
Third-gen NVLink for multi-GPU scaling.
312
TFLOPS FP16
624 TFLOPS with sparsity enabled.
19.5
TFLOPS FP64
Double-precision hardware for scientific and HPC workloads.
NVIDIA B200 on Denvr AI Cloud
1T Parameter Training
Scale across 8-GPU NVLink nodes for 1,440 GB of total VRAM and 1.8 TB/s per-GPU interconnect.
Proven Ecosystem
The most widely supported data center GPU in the ML stack. Full compatibility with PyTorch, TensorFlow, JAX, vLLM, and every major training and serving framework.
Distributed Training
Multi-node scaling with InfiniBand for distributed workloads that don't require H100-class throughput.
Managed Storage
High-performance Weka filesystem and local NVMe available. No external storage to provision for datasets, checkpoints, or model artifacts.
Configurations
Per-minute billing with on-demand and reserved options. All configurations available as bare metal, VM, or model endpoints.
Platform
GPUs
On-Demand
VRAM
vCPUs
Memory
Local Storage
Interconnect
NVIDIA A100 PCIe
4
40 GB
64
512 GB
2x 3.8TB NVMe
-
$1.15 / GPU