List your company
ResearchFormerly WizWand

Multimodal Reasoning on MMMU-Pro (mean@8 acc %)

51.1MMMU-Pro mean@8 AccRAPO
10.155220.785131.41542.0449May 10, 2026
Updated 5mo ago

Evaluation results

MethodLinks
RAPOBase Model=Qwen3-VL-8B-Instruct2026.0551.1
PAPOBase Model=Qwen3-VL-8B-Instruct2026.0550
VPPOBase Model=Qwen3-VL-8B-Instruct2026.0549.16
GRPOBase Model=Qwen3-VL-8B-Instruct2026.0548.5
BaseBase Model=Qwen3-VL-8B-Instruct2026.0537.59
RAPO_DBase Model=Qwen2.5-VL-7B-Instruct2026.0536.04
RAPO_GBase Model=Qwen2.5-VL-7B-Instruct2026.0536.01
R1-ShareVLBase Model=Qwen2.5-VL-7B-Instruct2026.0534.45
VL-RethinkerBase Model=Qwen2.5-VL-7B-Instruct2026.0534.1
RAPOBase Model=Qwen3-VL-2B-Instruct2026.0532.2
VPPOBase Model=Qwen3-VL-2B-Instruct2026.0531.73
PAPOBase Model=Qwen3-VL-2B-Instruct2026.0531.45
GRPOBase Model=Qwen3-VL-2B-Instruct2026.0530.61
RAPO_DBase Model=Qwen2-VL-7B-Instruct2026.0529.39
RAPO_GBase Model=Qwen2-VL-7B-Instruct2026.0528.22
MM-EurekaBase Model=Qwen2.5-VL-7B-Instruct2026.0526.82
BaseBase Model=Qwen3-VL-2B-Instruct2026.0524.44
TVCBase Model=Qwen2-VL-7B-Instruct2026.0523.34
BaseBase Model=Qwen2.5-VL-7B-Instruct2026.0521.22
MINT-CoTBase Model=Qwen2-VL-7B-Instruct2026.0517.63
BaseBase Model=Qwen2-VL-7B-Instruct2026.0511.73