Atirna 6cfb0974e3 fix: accurately price Claude benchmark runs (#177) ...

* fix: price result-less Claude runs correctly

* fix: reject Claude runs without reported costs

* fix: keep Claude pricing test scope focused

* fix: reject ambiguous Claude pricing

* test: cover every configured Claude pricing row

Sweep the full pricing table through the result-less fallback using the
token mix of a published agent-browser run, and pin the 1.25x cache
creation rate.

* no-mistakes(review): Fix Claude fallback usage pricing

* no-mistakes(review): Price US-only Claude inference

* no-mistakes(review): Charge Claude web searches

* no-mistakes(document): Document fallback web-search costs and format sources

* no-mistakes(lint): Fix map callback and format tests

2026-09-02 16:30:31 -07:00

..

grader.test.ts

feat: add browser benchmark (#8)

2026-03-25 17:38:59 -07:00

usage.test.ts

fix: accurately price Claude benchmark runs (#177)

2026-09-02 16:30:31 -07:00