Cost Tracking & Optimization
Track AI spend per request, per key, per model, and per team member. Set budget alerts, detect anomalies, and optimize costs automatically.
Features
Per-Request Costing
Every gateway request is priced based on input/output tokens and the provider's rate card. View exact cost breakdowns in the request log.
Budget Alerts
Set budget thresholds per virtual key or per project. Receive webhook and email notifications at 50%, 80%, and 100% of budget.
Anomaly Detection
ML-powered cost anomaly detection identifies unusual spending patterns — runaway loops, unexpected model upgrades, or denial-of-wallet attacks.
Cost Optimizer
Recommendations engine that suggests: cheaper equivalent models, prompt compression opportunities, caching candidates, and batch request consolidation.
| Metric | Granularity | Retention |
|---|---|---|
| Request cost | Per request | Plan-based (14–365 days) |
| Daily spend | Per key / per project | 90 days |
| Model cost comparison | Per model | 90 days |
| Budget utilization | Per key / per tenant | Current period |
| Cost anomalies | Per key | 30 days |