One row per agent: the total LLM spend, the average spend per execution, and the highest spend of any single execution. Every execution counts, whether or not the execution succeeded, because the spend was incurred either way.
Costs are computed from the token counts that each LLM call reports and the prices set on the model in its LLM profile. Set those prices to see costs in dollars. If the prices are left at zero, the report shows zero.
Cost: LLM spend per agent, from token counts and the profile's prices
Cost detail
Opening a row shows the agent's spend per published version. Each version has its own trend.
Cost detail: spend per published version, each with its own trend
