Skip to main content

Instance Types

NeuroHash offers a range of NVIDIA GPU instance types optimized for different AI workloads.

Available Instance Types

NVIDIA B200 (Blackwell)

The latest generation. 3x faster training and 15x faster inference compared to prior architectures.

ConfigGPU MemoryvCPUsRAMStoragePrice
1x HGX B200180 GB HBM3e30200 GiB1.5 TB$3.79/hr
8x HGX B2001,440 GB HBM3e2401,800 GiB25 TB$30.32/hr

Best for: Large language model training, next-gen AI research, high-throughput inference

NVIDIA H200

Near 2x the memory capacity of H100 with significantly higher bandwidth.

ConfigGPU MemoryvCPUsRAMStoragePrice
1x H200141 GB HBM3e26200 GiB1 TB$3.29/hr
8x H2001,128 GB HBM3e2081,800 GiB25 TB$26.32/hr

Best for: Memory-intensive models, large batch training, long-context LLMs

NVIDIA H100 SXM

The industry standard for LLM training and large-scale AI workloads.

ConfigGPU MemoryvCPUsRAMStoragePrice
1x H10080 GB HBM326200 GiB1 TB$2.69/hr
8x H100640 GB HBM32081,800 GiB25 TB$21.52/hr

Best for: LLM training, diffusion models, large-scale distributed training

NVIDIA A100 SXM

Proven, reliable GPU for training and inference at scale.

ConfigGPU MemoryvCPUsRAMStoragePrice
1x A10080 GB HBM2e30200 GiB1.4 TB$1.29/hr
2x A100160 GB HBM2e60400 GiB2.8 TB$2.58/hr
4x A100320 GB HBM2e120800 GiB5.6 TB$5.16/hr
8x A100640 GB HBM2e2401,800 GiB21 TB$10.32/hr

Best for: General training, fine-tuning, production inference

NVIDIA A10

Cost-effective option for inference and lighter workloads.

ConfigGPU MemoryvCPUsRAMStoragePrice
1x A1024 GB GDDR6X30200 GiB1.4 TB$0.75/hr

Best for: Inference, model serving, development, prototyping

NVIDIA GH200

Grace Hopper Superchip — CPU and GPU on a single module.

ConfigGPU MemoryvCPUsRAMStoragePrice
1x GH20096 GB HBM372480 GiB2 TB$2.49/hr

Best for: CPU-GPU tightly coupled workloads, memory-bound models

NVIDIA RTX A6000

Professional-grade GPU for research and visualization.

ConfigGPU MemoryvCPUsRAMStoragePrice
1x RTX A600048 GB GDDR614100 GiB512 GB$0.80/hr

Best for: Research, visualization, smaller training runs

Choosing the Right Instance

WorkloadRecommended GPUWhy
LLM training (7B-70B)8x H100 or 8x H200High memory, fast interconnect
LLM fine-tuning1x H100 or 1x A100Good balance of cost and performance
Inference serving1x A10 or 1x A100Cost-effective for serving
Diffusion models1x-4x A100Good memory/compute ratio
Development & prototyping1x A10Lowest cost
Cutting-edge research8x B200Maximum performance

Availability

Instance availability varies by region. When selecting an instance type in the launch modal, only regions with available capacity are shown. If your preferred GPU type is unavailable, try:

  • A different region
  • A different instance type
  • Checking back later (capacity changes frequently)