Sign inGet started
Available cloud GPU

Rent NVIDIA H100 GPUs on RemoteGPU

Launch H100 cloud GPUs for virtual machines, Kubernetes, and hosted Comfy workflows with hourly on-demand pricing and no long-term lock-in.

Get startedView all pricing
H100 GPU pricing

On-demand H100 plans

Compare H100 pricing across virtual machines, Kubernetes, and Hosted Comfy. Hourly prices are rounded for display.

Virtual Machine H100

Launch H100 virtual machines when you need direct guest access.

PlanGPU memoryCPUMemoryPrice / hourEst. / month
Virtual Machine H100 x180 GB8128 GiB$2.19/hr$1,598.70
Virtual Machine H100 x2160 GB16256 GiB$4.38/hr$3,197.40
Virtual Machine H100 x4320 GB32512 GiB$8.76/hr$6,394.80
Virtual Machine H100 x5400 GB40640 GiB$10.95/hr$7,993.50
Virtual Machine H100 x6480 GB48768 GiB$13.14/hr$9,592.20
Virtual Machine H100 x8640 GB641,024 GiB$17.52/hr$12,789.60

Kubernetes H100

Run H100-backed containers, jobs, services, and ingress on Kubernetes.

PlanGPU memoryCPUMemoryPrice / hourEst. / month
Kubernetes H100 x180 GB8128 GiB$2.19/hr$1,598.70
Kubernetes H100 x2160 GB16256 GiB$4.38/hr$3,197.40
Kubernetes H100 x4320 GB32512 GiB$8.76/hr$6,394.80
Kubernetes H100 x5400 GB40640 GiB$10.95/hr$7,993.50
Kubernetes H100 x6480 GB48768 GiB$13.14/hr$9,592.20
Kubernetes H100 x8640 GB641,024 GiB$17.52/hr$12,789.60

Hosted Comfy H100

Build and run Comfy workflows in a managed browser workspace.

PlanGPU memoryCPUMemoryPrice / hourEst. / month
Hosted Comfy H100 x180 GB8128 GiB$2.63/hr$1,918.44
Hosted Comfy H100 x2160 GB16256 GiB$5.26/hr$3,836.88
Hosted Comfy H100 x4320 GB32512 GiB$10.51/hr$7,673.76
Hosted Comfy H100 x5400 GB40640 GiB$13.14/hr$9,592.20
Hosted Comfy H100 x6480 GB48768 GiB$15.77/hr$11,510.64
Hosted Comfy H100 x8640 GB641,024 GiB$21.02/hr$15,347.52
Workloads

Where H100 fits

H100 capacity for production inference, batch jobs, fine-tuning, and image workloads.

LLM inference

Serve custom model endpoints and OpenAI-compatible text APIs on dedicated H100 capacity.

Image generation

Run diffusion and Comfy workflows with persistent storage and browser-based access.

Batch jobs

Schedule GPU containers and jobs on Kubernetes with clear hourly compute pricing.

Fine-tuning

Use 80 GB H100 memory per GPU for larger training and adaptation workloads.

FAQ

H100 rental questions

Can I rent one H100 at a time?

Yes. RemoteGPU publishes 1x, 2x, 4x, 5x, 6x, and 8x H100 plans where capacity is available.

Do H100 plans require a long-term contract?

No. Published H100 plans use hourly usage-based pricing unless a separate agreement says otherwise.

Which product surface should I use?

Use Token Factory for model endpoints, virtual machines for guest access, Kubernetes for container workloads, and Hosted Comfy for Comfy workflows.