Skip to main content
Cost policies control AI agent spending through daily budgets, per-request limits, and model tier enforcement. Prevent runaway costs from recursive agents, prompt injection attacks, or misconfigured workflows.

Budget — Daily $10

Cap total daily spending at $10 with alerts at 80% utilization.

Budget — Per Request $1

Limit individual request cost to $1 with pre-execution estimation.

Model Tier Enforcement

Restrict agents to cost-appropriate model tiers with automatic downgrade.

​
When to Use

Enable cost policies for any agent that makes LLM API calls. Daily budgets protect against runaway spending. Per-request limits prevent individual expensive operations. Model tier enforcement ensures agents use cost-appropriate models.
Combine daily budgets with per-request limits for defense-in-depth. A 10dailybudgetwitha10 daily budget with a 1 per-request limit prevents both gradual overspend and single expensive calls.