Solutions · Reserved tier
Hanzo Reserved tier
Capacity set aside for your organization — dedicated compute and inference for production traffic, or the whole stack on hardware you own — on an agreement sized to your load.
What it does for you
A dedicated compute cluster
GPU and CPU capacity held for your organization rather than drawn from the shared pool, so your traffic does not queue behind anyone else’s.Your models, hosted
Train and serve your own models on that capacity, beside the catalog, through the same API and the same key.Pay less for what can wait
A request can say how soon it needs its answer, and one that can wait is billed at a lower tariff. Reserved capacity carries the traffic that cannot.On your own hardware
Run the same open-source stack on-premises or air-gapped, with Zen weights served by Hanzo Engine. Moving between our cloud and yours is a deployment decision, not a migration.A custom SLA and people to call
The agreement carries a custom SLA, white-glove onboarding and 24/7 access to the engineers who run the platform.Capacity
On-demand GPUs today, per hour. Reserved capacity is priced in the agreement.
| GPUs | Memory | On demand |
|---|---|---|
| 1x H100 | 80 GB | $3.48 |
| 2x H100 | 160 GB | $6.96 |
| 4x H100 | 320 GB | $13.92 |
What it runs on
The products underneath, on one account, one key and one bill.
GPUs
Accelerators on demand, metered by the second.Machines
Sandboxes, metered machines and functions, each the organization’s own.Hanzo Engine
The inference engine, for serving models on machines you control.Enso
Routes each call to the model that should take it.Zen
Open-weight models you can download and run on your own hardware.Enterprise
Our cloud or yours: air-gapped deployment, SSO, audit logs and a person to call.More solutions
Put it to work
Start in the builder today, or talk to us about your organization.