Enterprise starts at a $2K/mo floor and adds in-region and on-prem deployment, SSO/SAML,
SCIM, an SLA, and solutions engineering. The policy engine is identical to the one the
cloud runs.
How overage works
Free includes 10,000 requests per month. Pro and Governance meter requests above the included volume at a flat per-request rate — you are billed for requests, never for a slice of the model spend that flows through them. Your provider bills (OpenAI, Anthropic, Google) remain yours, paid with your own provider keys; Spillway sits in front and governs them.Why not percentage-of-spend
Percentage-of-spend is common for observability middleware, but it’s the wrong incentive for a governance product. Spillway’s job is to shrink wasteful spend — block runaway budgets, route to the cheapest capable model, catch the anomaly before the invoice. If we charged a percentage of your bill, every dollar we saved you would cost us revenue. Flat tiers keep us aligned with the outcome you bought.Feature gating is enforced server-side — a call to a feature your plan doesn’t include
returns
402 tier_required (or plan_upgrade_required for quota/limit gates), the same
way a budget block does. See Errors.