Standard_ND96isr_H100_v5
Azure Virtual Machine: ND96isr_H100_v5 / ND96isr H100_v5 with 96 vCPUs and 1900 GiB of memory. Available in 24 regions starting from $71,773.60 per month. A -87.34% cheaper alternative is available.
| Standard_ND96isr_H100_v5 |
| Standard is recommended tier N – GPU enabled D – Training and inference scenarios for deep learning 96 – The number of vCPUs H100 – Accelerator Type i – Isolated size s – Premium Storage capable r – Remote direct memory access (RDMA) capable v5 – version |
| 96 |
| x64 |
| 1900 |
| V2 |
| 0 |
| 8 |
| NVIDIA H100 |
| 80 |
| 640 |
| 8 |
| yes |
| yes |
| 1023 GiB |
| 1024 GiB |
| 16 |
| yes |
| 0 |
| 80000 |
Regional Prices
Region Name | Region ID | Linux Price | Windows Price |
|---|---|---|---|
| East US | eastus | 98.3200 | 102.7360 |
| East US 2 | eastus2 | 98.3200 | 102.7360 |
| hidden-1 | hidden-1 | 98.3200 | 102.7360 |
| West US 2 | westus2 | 98.3200 | 102.7360 |
| hidden-2 | hidden-2 | 117.9680 | 122.3840 |
| hidden-3 | hidden-3 | 117.9840 | 122.4000 |
| hidden-4 | hidden-4 | 117.9840 | 122.4000 |
| hidden-5 | hidden-5 | 117.9840 | 122.4000 |
| hidden-6 | hidden-6 | 120.9340 | 125.3500 |
| hidden-7 | hidden-7 | 122.9000 | 127.3160 |
| hidden-8 | hidden-8 | 122.9000 | 127.3160 |
| hidden-10 | hidden-10 | 127.8160 | 132.2320 |
| hidden-11 | hidden-11 | 127.8160 | 132.2320 |
| hidden-12 | hidden-12 | 127.8160 | 132.2320 |
| hidden-13 | hidden-13 | 127.8160 | 132.2320 |
| hidden-9 | hidden-9 | 127.8160 | 132.2320 |
| hidden-14 | hidden-14 | 132.7320 | 137.1480 |
| hidden-15 | hidden-15 | 140.5980 | 145.0140 |
| hidden-16 | hidden-16 | 140.5980 | 145.0140 |
| hidden-17 | hidden-17 | 142.5640 | 146.9800 |
| hidden-18 | hidden-18 | 142.6000 | 147.0160 |
| hidden-19 | hidden-19 | 152.3960 | 156.8120 |
| hidden-20 | hidden-20 | 168.6720 | 173.0880 |
| hidden-21 | hidden-21 | 186.6080 | 191.0240 |
Best AI models you can run on this instance
Top open-weight chat models ranked by Intelligence Index that fit in Standard_ND96isr_H100_v5's 640 GB of GPU memory, assuming the model is sharded across all 8 GPUs. VRAM is estimated from the parameter count at the selected quantization, so treat it as guidance rather than a guarantee.
Chat Models
FP16 (full precision)
Model | Creator | Intelligence | Total params | Active params | ~VRAM | |
|---|---|---|---|---|---|---|
| DeepSeek V4 Flash |  DeepSeek | 40.3#28 | 158B | — | 379 GB | |
| MiniMax M2.7 |  MiniMax | 38.1#42 | 229B | — | 549 GB | |
| Qwen3.6 27B |  Alibaba | 37.1#51 | 27B | — | 65 GB | |
Similar Alternative VMs
Find Similar Instances In:
Microsoft Azure