List your company

Scientific Reasoning on GPQA Main (Accuracy, Tokens)

20.8AccuracyStatic
14.0415.79517.5519.305Apr 6, 2026

Evaluation results

MethodLinks
StaticBudget=40962026.0420.810,994
TABBudget (B)=10k2026.0419.46,415
LLM-Judge Multi-TurnSelection Strategy=Multi-Turn2026.0418.15,712
StaticBudget=20482026.0417.45,910
LLM-Judge IndividualSelection Strategy=Individual2026.0417.45,940
TABBudget (B)=8k2026.0417.24,843
StaticBudget=10242026.0416.73,929
TABBudget (B)=5k2026.0416.52,971
TABBudget (B)=3k2026.0415.62,258
StaticBudget=5122026.0414.72,544
StaticBudget=2562026.0414.31,822