TRACEBURN / SYNTHETIC DEMO

Where did this run spend?

Inspect the calls. Test a change. Check that the task still succeeds.

Estimated model cost$0.0360
Model calls3
Findings to investigate1

Costs are estimates from recorded metadata. Cost changes alone do not establish equivalent task quality.

What to investigate

Repeated model requests

2 linked calls | high confidence

Inspect the linked calls locally to check whether the requests and answers repeat.

Call 2, Call 3

Try next: Try reusing a result or moving the call out of the loop when its inputs are stable.

When this may be expected

Repeated sampling can be intentional. Check task outcomes before removing a call.

Potential savings: $0.0120. Estimate only; findings may overlap and should not be added together.

Calls and recorded usage

Call numbers are local to this report. Input excludes cache reads and writes, which have separate columns. Missing usage is shown as unknown.

CallParentKindStatusTime (ms)InputCache readCache writeOutputEst. cost
Call 1rootagentok4,000.0unknownunknownunknownunknownn/a
Call 2Call 1llmok1,000.02,40000160$0.0120
Call 3Call 1llmok1,000.02,40000160$0.0120
Call 4Call 1llmok1,000.02,40000160$0.0120