Token Usage & Unit Economics - OpsBeacon

Unit economics for AI

Raw token spend is hard to act on. Unit economics — cost per API call, cost per successful completion, cost per user request — give you a number you can improve. OpsBeacon calculates unit economics for each AI project where usage data is available.

How to use it

@Beacon what's the cost per API call for the summarization project?
@Beacon show me unit economics for all AI projects
@Beacon which project has the worst cost per completion?

Key metrics

Cost per 1K tokens — blended input + output token cost
Cost per API call — average spend per request
Input/output ratio — prompt tokens vs completion tokens (high input ratios often indicate prompt inefficiency)
Average completion length — longer completions cost more; are they justified?

⌘I