Clear costs before the GPU starts.
Pick the platform access you need. Then pay the visible provider rate for the infrastructure you use. Winterr fees stay separate, from quote to statement.
Every charge stays legible
- Provider infrastructure
- Source rate
- Winterr platform
- Fixed plan
- Managed products
- Explicit fee
- Expected total
- Quoted first
Choose how you build
Three self-serve plans. Upgrade when resource and team complexity grows.
USD · billed monthlyProvider usage billed separately
Explore
Free
$0
/ month
For evaluating Winterr with one active GPU resource.
Everything you need to explore
- 1 active GPU resource
- Manual quotas
- Limited history
- Public Templates
Build
Most popularPro
$49
/ month
For developers shipping repeatable AI workloads.
Everything you need to build
- 3 active GPU resources
- Transparent source rates
- Private Templates
- API, SDK, CLI, and MCP access
Scale
Team
$299
/ month
For platform teams that need shared control.
Everything you need to scale
- 20 active GPU resources
- Role-based access
- Shared wallet
- Team quotas and audit
Compare at a glance
Plan limits and controls, grouped around the way you work.
| Feature | Free | Pro | Team |
|---|---|---|---|
| Capacity | |||
| Monthly platform fee | $0 | $49 | $299 |
| Active GPU resources | 1 | 3 | 20 |
| Templates | Public | Public + private | Public + private |
| Developer access | |||
| Dashboard | Included | Included | Included |
| API, SDK, CLI, and MCP | — | Included | Included |
| Transparent source rates | Included | Included | Included |
| Team controls | |||
| Role-based access | — | — | Included |
| Shared wallet | — | — | Included |
| Team quotas and audit | — | — | Included |
Usage follows the value delivered
Compute is measured as infrastructure usage. Higher-level products add only the runtime or orchestration work they use.
- Raw Compute
- Provider compute, storage, and network at the visible source rate.
- Jobs and Services
- Underlying resources plus an explicit orchestration fee when used.
- Sandboxes
- CPU, memory, storage, network, and isolation-backend runtime.
- Benchmarks
- Job or Sandbox compute, retained artifacts, and evaluation-case processing.
- Training
- Provider compute and storage plus an explicit managed-run orchestration fee.
- Reinforcement learning
- Rollout, Sandbox, trainer, storage, and an explicit workflow fee.
No surprise spend
Money controls run before provider infrastructure is created.
- QuoteLock provider terms and the customer price snapshot.
- HoldReserve wallet credit for the configured risk window.
- LimitEnforce resource, daily spend, and operation-level budgets.
- SettleRecord usage, release unused holds, and preserve the ledger trail.
Pricing questions, answered
The details buyers usually need before they run a first workload.
Is compute included?
No. The plan covers Winterr platform access and plan limits. Provider compute, storage, and network usage are billed separately at the visible source rate.
Does Winterr mark up compute?
Winterr does not hide a markup inside the provider rate. The source price and Winterr software fees remain separate line items.
When do managed fees apply?
Jobs, Services, Sandboxes, Benchmarks, Training, and reinforcement-learning workflows can add explicit orchestration, runtime, case-processing, or workflow fees. The expected charge shape is shown before a long-running operation starts.
How do upgrades work?
Move from Free to Pro when you need private Templates and programmatic access. Move to Team when you need more active GPU resources, role-based access, a shared wallet, and team policy controls.
What comes after Team?
Enterprise contracts can set a larger resource quota and add private or BYOC deployment, support and SLO tiers, and commercial controls.
How is spend controlled?
Winterr quotes the operation, reserves wallet credit, enforces resource and spend limits, then settles actual usage and releases unused holds.
Start small. Keep every cost visible.
Open the dashboard, fund a wallet, and review a quote before you provision.