Plan details are temporarily unavailable. Please check back shortly or contact us.
How billing works
- Start with credit
Free buys credit from $5. The Pro pack is $20 of credit plus higher caps for 30 days. Enterprise is a prepaid annual commit.
- Run your workload
Train, deploy or spin up agents on any GPU in the live catalog. Rates are identical on every plan; only the caps change.
- Meter while it runs
GPU-hours from instance boot, sandbox awake-hours, GiB-months and traces are drawn from credit as they happen.
- Stop and it stops
Cancel before boot and pay nothing. Deployments scale to zero. Work stops at a $0 balance unless auto-reload is on.
Plan limits, side by side
Plan limits load from the control plane; please check back shortly.
Usage pricing
Loading live GPU pricing…
Stopped replicas cost nothing. Model bytes stay until you delete them.
Billed per minute while awake; paused and ended sessions cost nothing. Egress is charged only in audited_open mode. Model calls bill through your Veri deployment or your own provider.
The 512 GiB volume allowance is the same on every plan.
Opt in per deployment (off by default); full request and response bodies, metered from credit at month end.
Per-GPU throughput is illustrative until measured on our serving stack. Price per GPU-minute is the live H100 rate divided by 60.
We'll map your workload to the right GPUs, caps and billing.