Token Usage Tracking
Know exactly what your AI agents cost and where every token goes.
Visibility into AI spend
Running a fleet of AI agents means managing API costs across multiple models and providers. Token Usage Tracking gives you a single dashboard to monitor consumption across your entire organization — broken down by agent, model, and time period.
Every API call is logged with input tokens, output tokens, cache hit rate, and estimated cost. You can see which agents are the most expensive, which tasks consume the most context, and where prompt caching is saving you money.
Set per-agent budgets and get alerts before an agent exceeds its allocation. Token usage data is retained for 90 days and can be exported for financial reporting.
What we track
| Metric | Description |
|---|---|
| Input tokens | Tokens sent to the model in prompts and context |
| Output tokens | Tokens generated by the model in responses |
| Cache hits | Tokens served from prompt cache, reducing cost |
| Estimated cost | Dollar cost based on model pricing tiers |
Budget controls
Set daily or monthly token budgets per agent. When an agent approaches its limit, you get a notification. When it hits the limit, the agent pauses and waits for approval to continue. This prevents runaway costs from long-running tasks or retry loops — you stay in control without micromanaging.