List your company

Aggregate Cognitive Performance on Average

76.6Human-Model Similarity (1-nWD)TASKPR
-3.06417.61838.358.982May 25, 2026

Evaluation results

MethodLinks
TASKPRModel Backbone=Llama 3 8B2026.0576.6
TASKPRModel Backbone=Claude Opus 4.62026.0560.3
COMPACTORModel Backbone=Claude Opus 4.62026.056.9
COMPACTORModel Backbone=Llama 3 8B2026.055.4
MEMPRModel Backbone=Llama 3 8B2026.051.6
HUMPRModel Backbone=Llama 3 8B2026.051
HUMPRModel Backbone=Claude Opus 4.62026.050.4
MEMPRModel Backbone=Claude Opus 4.62026.050