Terminal X eval agent matches senior investor judgment 88.6% with GPT-6.1

Terminal X's Eval Agent grades outputs against a 54-criterion rubric using GPT-6.1 Sol, catching 25% more deviations from a fund's mandate compared to traditional methods.