List your company

Multimodal Reasoning on VisualPuzzles (OOD)

33.9Overall AccuracyAMVL
27.493629.156830.8232.4832Jul 1, 2026

Evaluation results

MethodLinks
AMVLSize=7B2026.0733.932.4427.9652.528.2330.77
Deepeyes2026.0732.9637.7927.014126.7927.01
Qwen2.5-VL-7BSize=7B2026.0732.7137.0221.847.526.3221.8
Pangea-7BSize=7B2026.0731.332.423.738.528.732.5
LLaVA-OneVision-72BSize=72B2026.0730.834.726.53727.328.7
LVR2026.0727.7428.6323.223628.2324.12