--- title: "Pricing, Usage, and Limits" description: "How to think about Norcaster plans, usage, managed vs self-hosted pricing, budget gates, and enterprise limits." canonical: https://www.norcaster.com/docs/pricing-limits --- Pricing # Pricing, usage, and limits For **Norcaster Boundary** plan names, limits, and enforcement notes (Starter, Team, Enterprise), see [Boundary pricing](https://www.norcaster.com/pricing) and [Boundary pricing (docs)](https://www.norcaster.com/docs/boundary-pricing) — scoped to Boundary, not the entire platform catalog. This doc explains how the wider Norcaster product thinks about usage, hosting, budget gates, and enterprise limits so teams know what to ask before scaling. ## What counts as usage - Number of deployments and provider-backed inference targets. - Workflow executions, model calls, retrieval queries, and tool calls. - Audit exports, trace retention, eval runs, and observability volume. - Team, SSO, compliance, and managed hosting requirements. ## Managed vs self-hosted cost model ### Managed Norcaster Norcaster operates the control plane. You pay for platform access and usage while still paying model providers directly where applicable. ### Self-hosted Norcaster You operate infrastructure, storage, and runtime services. This can reduce data movement but shifts operational responsibility to your team. ## Budget gates Budget gates are policy controls, not billing invoices. They help prevent runaway spend by limiting request rate, compute seconds, retry amplification, or other runtime ceilings before execution continues. [Set up budget control examples →](https://www.norcaster.com/docs/budget-control) ## Common questions ### Do provider model costs come from Norcaster? Usually no. If you connect your own Anthropic, OpenAI, Azure OpenAI, Bedrock, Groq, Together, or Voyage AI account, provider usage is governed by that provider relationship. ### Can I export my audit trail? Yes. Norcaster supports enforcement-event exports and signed audit exports when signing is configured.