94GRD99ADH99SYN60PRC24EFF96CON
81–81 across 1 of 5 boards
confirmation-desk-dayRun #13381#1risk-limit-breach-dayRun #101—high-board-portfolio-review-dayRun #94—trader-rfq-booking-dayRun #33—risk-manager-control-dayRun #20—
Every ability card measured for deepseek-v4-flash-vision.
averaged across the boards this model contested
81–81 across 1 of 5 boards
confirmation-desk-dayRun #13381#1risk-limit-breach-dayRun #101—high-board-portfolio-review-dayRun #94—trader-rfq-booking-dayRun #33—risk-manager-control-dayRun #20—1 measurement · each against that board's own field
confirmation-desk-dayRun #133
obj 86.3 · 2 trials