IDP
Loading…
AI token usage costs, budget tracking, and optimization
| Agent | Model | Input Tokens | Output Tokens | Sessions | Cost |
|---|---|---|---|---|---|
| Commitment Agent | GPT-4o | 187K | 98K | 623 | $1,142 |
| Reporting Agent | GPT-4o | 62K | 41K | 289 | $487 |
| Escalation Handler | GPT-4o | 38K | 19K | 147 | $271 |
| Onboarding Agent | GPT-4o-mini | 25K | 12K | 188 | $47 |
| Total | 312K | 170K | 1,247 | $1,947 | |
Last 14 days
Monthly spend at 91% of budget ($3,120 projected vs $3,500 budget)
Reporting Agent under budget — consider upgrading model for better quality
Savings: Escalation Handler could use GPT-4o-mini for routine triage — est. $89/mo savings
Switch Onboarding Agent to GPT-4o-mini — 78% cost reduction with minimal quality impact for simple Q&A tasks
Implement response caching for Commitment Agent — repeated queries account for ~30% of tokens
Add topic-level token budgets to prevent runaway sessions