Quotas & plan gating

Every plan includes a monthly request quota: Starter 5M, Pro 50M, and Scale 500M requests per month. This is a soft cap, not a wall — hitting 100% doesn’t stop your traffic. You get alerts and a dashboard banner, and Detent keeps serving requests as normal while overage is tracked. The quota is there to keep you informed about usage, not to interrupt production traffic at an inconvenient moment.

Above the quota sits an optional anti-abuse hard ceiling, configured separately from the quota itself. This one does enforce: crossing it returns 429 with code monthly_hard_cap instead of a normal verdict. It exists as a backstop against runaway or abusive usage, not as a billing mechanism — most accounts will never hit it.

Algorithm gating

Which algorithms you can use is also plan-based. Starter includes sliding_window and fixed_window. Pro and Scale add concurrent_limit for in-flight limiting. Requesting an algorithm that isn’t included in your plan returns 403 with code algorithm_not_on_plan — the request is rejected outright rather than silently falling back to a different algorithm.

Both of these are 4xx/429 conditions with a stable code field; see Machine error codes for the full list and how to handle them in your client.