Instances are available as virtual machines and bare metal, providing flexibility and high performance.
Shape |
GPUs |
Architecture |
Network |
GPU Price Per Hour ** |
Large scale-out AI training, data analytics, and HPC | ||||
---|---|---|---|---|
BM.GPU.H100.8 |
8x NVIDIA H100 80GB Tensor Core |
Hopper |
8x2x200 Gb/sec |
|
BM.GPU.A100-v2.8 |
8x NVIDIA A100 80GB Tensor Core |
Ampere |
8x2x100 Gb/sec RDMA* |
|
BM.GPU4.8 |
8x NVIDIA A100 40GB Tensor Core |
Ampere |
8x2x100 Gb/sec RDMA* |
|
Small AI training, inference, streaming, gaming, and virtual desktop infrastructure | ||||
VM.GPU.A10.1 (new) |
1x NVIDIA A10 Tensor Core |
Ampere |
24 Gb/sec |
|
VM.GPU.A10.2 (new) |
2x NVIDIA A10 Tensor Core |
Ampere |
48 Gb/sec |
|
BM.GPU.A10.4 |
4x NVIDIA A10 Tensor Core |
Ampere |
2x50 Gb/sec |
|
VM.GPU3.1 |
1x NVIDIA V100 Tensor Core |
Volta |
4 Gb/sec |
|
VM.GPU3.2 |
2x NVIDIA V100 Tensor Core |
Volta |
8 Gb/sec |
|
VM.GPU3.4 |
4x NVIDIA V100 Tensor Core |
Volta |
25 Gb/sec |
|
BM.GPU3.8 |
8x NVIDIA V100 Tensor Core |
Volta |
2x 25 Gb/sec |
|
VM.GPU2.1 |
1x NVIDIA P100 |
Pascal |
8 Gb/sec |
|
BM.GPU2.2 |
2x NVIDIA P100 |
Pascal |
2x 25 Gb/sec |