Hyperbolic On-Demand GPU Compute
Hyperbolic On-Demand GPU Compute rents NVIDIA H100, H200 and B200 instances by the hour, or as reserved clusters, for AI training, fine-tuning and inference.
Find alternatives to Hyperbolic On-Demand GPU Compute
What it does
- GPU cloud
- Model training
- Model inference
Features 5
- On-demand GPU instancesInstances for H100, H200 and B200 GPUs can be launched in minutes with no commitment.
- Multi-node GPU clustersClustered allocation scales from one node to 1000+ GPUs and deploys in minutes for distributed training.
- Bare-metal or VM setupUsers choose between fast VMs and bare-metal performance when setting up an instance or cluster.
- InfiniBand or Ethernet interconnectUsers pick InfiniBand or Ethernet as the cluster interconnect when configuring capacity.
- Network isolation between sessionsOn-demand workloads are isolated from other users with network isolation and instance cleanup between sessions.
Sources
- Pricing
-
Pricing page: "Rent GPUs starting at $3.19/GPU/hr, cutting compute costs for training and inference."
- Description
-
Official documentation · 10 Nov 2026
Product page: "Access NVIDIA H100, H200, and B200 GPUs for AI training, fine-tuning, and inference."
- What it does
-
Official documentation · 10 Nov 2026
Product page: "Access NVIDIA H100, H200, and B200 GPUs for AI training, fine-tuning, and inference."
- Sold within
-
Sold on its own with published plans.
- Name
-
Official documentation · 10 Nov 2026
Name as published on the product page.
- URL
-
Official documentation · 10 Nov 2026
Product page, read 2026-11-10.
- Deployment
-
Official documentation · 10 Nov 2026
Product page: "Launch on-demand GPU instances in minutes or reserve dedicated GPU clusters for sustained workloads."
- Interfaces
-
Official documentation · 10 Nov 2026
Product page: "Launch on-demand GPU instances in minutes or reserve dedicated GPU clusters for sustained workloads."
Also from Hyperbolic
-
Hyperbolic Inference Model API
Serverless pay-per-token inference API serving eleven open models from labs including DeepSeek, Z.ai, Moonshot AI, NVIDIA and Xiaomi through one OpenAI-compatible endpoint on Hyperbolic's Flux platform, with no GPUs to provision.
-
Hyperbolic Managed Kubernetes Infrastructure service
Managed GPU Kubernetes clusters created from the Hyperbolic console, with a hosted control plane, GPU worker nodes, NVIDIA GPU and network operators and cluster health and GPU metrics, for containerized training, fine-tuning and inference; in beta, with workers billed at the on-demand GPU instance rates.
Open Hyperbolic in the directory
Something wrong here? Send a correction — quote this product id: hyperbolic-on-demand-gpu-compute.