Blackwell GPU
clusters.
Explore Helios’s reserved NVIDIA GPU offering. Choose a configuration, review the published terms and talk to our team about capacity for your workload.
Capacity and deployment dates are confirmed with your request.
Start with the hardware.
Compare the GPU options in our published reserved schedule. Select a term to see its per-GPU hourly rate.
Selected configuration
NVIDIA B300
- GPU memory
- 288GB
- Published reserved rate
- $4.37–$4.59USD / GPU-hour
Specify your GPU count, node configuration and workload when requesting capacity. The rate shown is tied to a 3-year reservation; it is not an on-demand or per-token price.
See the B300 reference build and BOM Request this configurationYour GPU choice and reservation term carry into your request. We’ll confirm capacity and timing with you.
Define the cluster around the job.
A GPU choice is the starting point. Include these requirements so your capacity request can be scoped as a complete system.
GPU cluster reservation checklist- Compute
- GPU type, quantity, model size and expected run duration.
- Network & storage
- Multi-node communication, dataset size and checkpoint requirements.
- Software & access
- Your framework, orchestration preference and access requirements.
- Deployment
- Preferred location, start date, reservation term and support expectations.
Bring us your workload.
We’ll start with configuration, capacity and timing.