Plot benchmark performance against inference cost
#40
by jfbrown95 - opened
Show Youden’s J against captured run cost with a highlighted Pareto frontier for each benchmark. Replace Liquid d1’s current rows with the complete paid OpenRouter rerun, including API-reported costs and public predictions. Preserve question cutoffs and failure scoring.
jfbrown95 changed pull request status to merged