Rent NVIDIA H100 NVL GPU by the hour
The NVIDIA H100 NVL is a Hopper-architecture datacenter GPU equipped with 94 GB of VRAM and compute capability 9.0. It delivers about 48.3 TFLOPS of single-precision performance and about 3381 GB/s of memory bandwidth within a 400 W thermal envelope. Rent the H100 NVL on ONIE with per-hour billing, SSH access, and ready templates for PyTorch/CUDA, vLLM, Ollama, and ComfyUI.
Specifications
VRAM per GPU
Architecture
Compute capability
FP32 compute
Memory bandwidth
Power limit
Values reported by the infrastructure provider for machines currently listed; real performance varies by machine.
Available configurations
Live stock and prices, updated continuously. The final price is confirmed before you order.
Loading live prices…
What the H100 NVL is good for
Large Language Model Inference
Deploy high-capacity language models with vLLM or Ollama using 94 GB of VRAM and about 3381 GB/s of memory bandwidth for fast token generation.
Deep Learning Fine-Tuning
Train and fine-tune neural networks in PyTorch using Hopper architecture features and compute capability 9.0.
High-Resolution Image Generation
Run complex ComfyUI workflows and large diffusion pipelines in memory without offloading to system RAM.
Scientific Computing and Simulation
Execute memory-intensive datacenter workloads that benefit from about 48.3 TFLOPS of single-precision compute.
Is it the right GPU for you?
Choose the H100 NVL when your AI or compute workloads require more memory than standard datacenter cards provide, up to 94 GB of VRAM. A smaller workstation or consumer GPU is a better choice for lightweight inference and development, while a multi-GPU setup or a larger memory tier is necessary if your models exceed 94 GB.
Compare with similar GPUs
Frequently asked questions
Is the H100 NVL suitable for LLM inference and fine-tuning?
Yes, its 94 GB of VRAM and about 3381 GB/s of memory bandwidth make it well suited for serving and fine-tuning large language models that do not fit on smaller datacenter GPUs.
How does the H100 NVL compare to standard H100 variants?
The H100 NVL offers an expanded 94 GB VRAM capacity and a 400 W power envelope, providing extra memory headroom compared to standard Hopper datacenter configurations.
What software templates are available on ONIE for this GPU?
You can deploy the H100 NVL using pre-configured templates for PyTorch/CUDA, vLLM, Ollama, and ComfyUI, all supporting compute capability 9.0.
How do billing and access work when renting the H100 NVL?
Instances on ONIE use per-hour billing and provide direct SSH access so you can manage your environment and workflows immediately.
Rent the H100 NVL by the hour
Create an account, choose a configuration and connect over SSH. Billing stops when you delete the GPU.
Rent H100 NVL