Usage and limits

Every key carries a budget and a rate limit. Both are enforced in the router before a request reaches a provider, so an overrun never becomes a surprise invoice.

Budgets

Set a daily or monthly cap in dollars. At 50, 80 and 100 percent the key fires an alert to email or a webhook. At 100 percent you choose: hard-stop with a 402, or downgrade to the fast tier and keep serving.

Rate limits

Requests per minute and tokens per minute, per key. Prism's own limits are additive to each provider's; when a provider limit is the binding one, the router spreads load across providers before it returns a 429.

The usage view

Spend, tokens and requests by model, key, policy and day. Every number in the dashboard is also a row in the export: GET /v1/usage?from=…&to=…&group=model returns CSV or JSON.

Retention

Plan	History
Free	7 days
Team	90 days
Enterprise	Custom, up to 3 years