List your company
ResearchFormerly WizWand

Reasoning on AIME 2025 (Accuracy)

49.58AIME 2025 AccuracyCoPD
-1.983211.403424.7938.1766Sep 25, 2025Oct 31, 2025Dec 6, 2025Jan 11, 2026Feb 16, 2026Mar 24, 2026Apr 29, 2026
Updated 6mo ago

Evaluation results

MethodLinks
CoPDTraining Strategy=Co-Evolving Policy Distillation2026.0449.58
CoPDTraining Strategy=CoPD2026.0449.38
Text-ExpertTraining Strategy=Text-Specific Expert2026.0448.33
Text-ExpertTraining Strategy=Text-Expert2026.0448.33
Image-ExpertTraining Strategy=Image-Specific Expert2026.0447.5
Image-ExpertTraining Strategy=Image-Expert2026.0447.5
Video-ExpertTraining Strategy=Video-Expert2026.0447.29
BaseTraining Strategy=Base Model2026.0446.88
BaseTraining Strategy=Base2026.0446.88
MOPDTraining Strategy=MOPD2026.0446.04
Mixed RLVRTraining Strategy=Mixed RLVR2026.0445.62
OPD T→VTraining Strategy=Static On-policy Policy Distillation (Text to Image)2026.0445.42
Mixed RLVRTraining Strategy=Joint optimization on combined budget2026.0444.58
OPD V→TTraining Strategy=Static On-policy Policy Distillation (Image to Text)2026.0443.54
VanillaModel=Llama2025.0943.3
OjaKVModel=Llama2025.0913
Eigen-NModel=Llama2025.090
StaticPCAModel=Llama2025.090
PaluModel=Llama2025.090