Skip to content
AI usage dashboard

synthetic-cost-estimated._expectations.md (synthetic)

All files are synthetic, not native session extracts. Test estimated usage.cost.total: map positive values, leave zero unknown because it cannot be distinguished from missing prices. Shape: Pi session JSONL v3.

Scenario and expectations

Three lines: session syn-sess-cost; assistant syn-c-1, usage 100/50/0/0/150, cost.total=0.005; assistant syn-c-2, usage 5/5/0/0/10, cost.total=0.

  • syn-c-1: cost_amount_minor=0.005×1_000_000=5000, USD, kind=estimated; absent price_version remains NULL.
  • syn-c-2: cost.total=0 leaves cost_amount_minor/cost_currency/cost_kind NULL.
  • 2026-01-05 summary: call_count=2, input_total_known=105 (100+5), output_total_known=55 (50+5), total_tokens_known=160 (150+10).