About Hyperbolic
Hyperbolic is a cloud platform that offers GPU compute and inference for AI builders. It provides on-demand H100, H200 and B200 GPUs along with managed clusters.
What it does
Teams can launch GPU instances in minutes or reserve dedicated clusters for longer workloads. Developers can also run leading models through an OpenAI-compatible API with token-based billing. Managed Kubernetes with a hosted control plane is available for teams that run clusters of their own.
Key features
- On-demand H100, H200 and B200 GPUs
- Reserved multi-node GPU clusters
- Managed Kubernetes with a hosted control plane
- OpenAI-compatible inference API
- Token-based billing for models
Who it's for
- Startups that need GPUs without long contracts
- Teams running training or fine-tuning jobs
- Developers who want model inference without managing GPUs
Pricing and good to know
GPU instances and reserved clusters have separate pricing, which the site links to. Inference is billed by token through your account. Contact sales for large or long-term capacity, and compare GPU types against your workload before you reserve hardware.
Highlights
- Launch on-demand H100, H200 and B200 GPUs
- Reserve multi-node GPU clusters
- Run managed Kubernetes clusters
- Call models through an OpenAI-compatible API
- Pay per token for inference
