Every plan runs the same runtime, in the same regions, on the same hardware. What changes is how much capacity comes included, and how much of our time comes with it.
Developer
$0 per month
For evaluating the runtime with a real endpoint.
Team
$490 per month + usage
For production workloads run by a small platform team.
Enterprise
Custom
For regulated or high-volume environments.
Usage beyond the included hours is billed per GPU-hour by tier — see the estimator and the full comparison. Prices exclude taxes.
Pick a hardware tier and how many GPU-hours you expect in a month. The estimate shows the plan that fits and how the number is built — no hidden fees, no surprise tier.
Recommended plan
Team
$490/ month
Estimate only. Egress and taxes not included. Enterprise pricing is committed and negotiated.
| Feature | Developer | Team | Enterprise |
|---|---|---|---|
| Core runtime | |||
| Included capacity | 50 GPU-h / month | 500 GPU-h / month | Reserved, committed |
| Usage beyond included | — | Per GPU-hour, per tier | Committed rate |
| Dynamic batching | Yes | Yes | Yes |
| Autoscaling & scale-to-zero | Yes | Yes + warm pools | Yes + warm pools |
| Bring your own container | Yes | Yes | Yes |
| Regions & hardware | |||
| Regions | 2 | All 24 | All 24 + private regions |
| Hardware tiers | L40S, CPU | H200, Gaudi 3, L40S, CPU | All + reserved H200 |
| Rollouts | Direct | Canary with auto-pause | Canary with auto-pause |
| Data residency pinning | — | — | Yes |
| Admin & security | |||
| Team members | 1 | Up to 10 | Unlimited |
| Single sign-on | — | — | SAML / OIDC |
| Audit logs | — | 30 days | 1 year, exportable |
| DPA & security review | — | Standard DPA | Custom DPA, questionnaire |
| Support | |||
| Channel | Community | Email, 1 business day | Named engineer, shared channel |
| SLA | — | 99.9 % | 99.99 % |
| Migration assistance | — | Guides | Hands-on |
The same runtime your engineers evaluated on the free tier, with the paperwork and guarantees your security and finance teams need. No feature is held hostage behind the sales call.
Committed H200 capacity at a fixed rate, in the regions you name.
Pin workloads to jurisdictions; run on dedicated capacity with VPC peering.
SAML / OIDC, role-based access, one year of exportable audit logs.
DPA, sub-processor list, security questionnaire, PO invoicing.
A named engineer, a shared channel and a 99.99 % SLA with credits.
Anything missing? Ask an engineer — we answer within a business day.
Per GPU-second of allocated capacity plus egress, metered per region. Autoscaling to zero means idle deployments cost nothing; warm pools are billed as allocated capacity. Included hours reset monthly.
Nothing breaks. Team keeps running and bills the extra GPU-hours at the tier rate shown above. Developer pauses new capacity and emails you; upgrade in one click.
Yes. Monthly plans stop at the end of the billing period; annual plans run to the end of the term. Your deployment files stay yours and export with terrane export.
Annual billing saves 15 %. Startups under two years old and open-source projects can apply for credits from the dashboard; enterprise committed use is priced on volume.
No. Terrane is the runtime. You describe services in one file and the platform schedules them on managed capacity. There is no cluster to create, upgrade or right-size.
Requests are processed in the region that serves them and are not persisted beyond logs you control. Enterprise plans can pin deployments to specific jurisdictions.
A DPA, a security questionnaire and invoicing with a PO are standard on Enterprise. We can complete your vendor form and provide our sub-processor list on request.
Two regions, 50 GPU-hours and a real endpoint. Upgrade when your traffic asks for it, not before.