| Flavour Name | GPUs | vCPUs | RAM | Hourly | Monthly | 6 Monthly 5% Off | 12 Monthly 10% Off | |
|---|---|---|---|---|---|---|---|---|
| N.H200.16.128 | 1x | 16 | 128 | --- | $3,298 | $18,797 | $35,615 | |
| N.H200.24.192 | 1x | 24 | 192 | --- | $3,558 | $20,283 | $38,431 | |
| N.H200.28.282 | 1x | 28 | 282 | --- | $3,920 | $22,342 | $42,331 | |
| N.H200.48.384 | 2x | 48 | 384 | --- | $7,116 | $40,563 | $76,856 | |
| N.H200.56.564 | 2x | 56 | 564 | --- | $7,839 | $44,680 | $84,656 | |
| N.H200.96.768 | 4x | 96 | 768 | --- | $14,232 | $81,123 | $153,706 |
See how much your GPU setup could cost across AceCloud, AWS, Azure and Google Cloud. Choose a GPU and quantity to compare monthly estimates.
Compare GPU CostsAceCloud offers NVIDIA H200 NVL cloud GPU instances in its Atlanta, USA region with pricing listed in USD. Available pricing varies by GPU configuration, vCPU, RAM and billing term. Check the pricing table above for the current monthly, 6-month and 12-month rates.
These prices are for NVIDIA H200 NVL cloud GPU instances on AceCloud, not for purchasing a physical NVIDIA H200 GPU or server.
NVIDIA H200 provides 141GB of HBM3e GPU memory with up to 4.8 TB/s memory bandwidth. The additional memory capacity and bandwidth make it particularly useful for large language models, long-context inference, generative AI and memory-intensive HPC workloads.
Yes. AceCloud provides monthly, 6-month and 12-month pricing options for H200 NVL cloud GPU instances. Longer-term plans can provide a lower effective monthly cost compared with standard monthly pricing.
NVIDIA H200 GPUs are designed for demanding workloads such as large language model inference, model training and fine-tuning, generative AI, RAG pipelines, multimodal AI and high-performance computing. Its 141GB HBM3e memory is especially useful for workloads constrained by GPU memory capacity or bandwidth.
Yes. NVIDIA H200 is designed for large AI models and memory-intensive inference workloads. Its 141GB HBM3e memory allows larger models, KV caches, context windows and batch sizes to remain in GPU memory, which can improve throughput for LLM workloads.
AceCloud offers multi-GPU H200 configurations for workloads that need additional compute and GPU memory capacity. Multi-GPU instances can be used for larger AI training, inference and HPC workloads that can scale across GPUs.