List your company

Compositional Reasoning on WhatsUp

51.3AccuracyReasonCLIP
40.27643.1384648.862Jun 25, 2026

Evaluation results

MethodLinks
ReasonCLIPViT=B/32, Resolution=224, Data=+58M2026.0651.3
ReasonSigLIPViT=So/14, Resolution=384, Data=+58M2026.0649.7
MetaCLIPViT=L/14, Resolution=224, Data=2.5B2026.0648.6
SigLIPViT=So/14, Resolution=384, Data=10.0B2026.0647.6
SigLIP2ViT=G/16, Resolution=384, Data=10.0B2026.0646.9
MetaCLIPViT=B/32, Resolution=224, Data=0.4B2026.0646.5
Eva-CLIP-02ViT=L/14, Resolution=224, Data=2.0B2026.0646.3
ReasonCLIPViT=L/14, Resolution=336, Data=+58M2026.0646.3
ReasonCLIPViT=L/14, Resolution=224, Data=+58M2026.0646
Long-CLIPViT=L/14, Resolution=224, Data=0.4B2026.0645.6
SigLIP2ViT=L/16, Resolution=384, Data=10.0B2026.0645.5
ReasonSigLIP2ViT=G/16, Resolution=384, Data=+58M2026.0645.3
ReasonCLIP + READViT=B/32, Resolution=224, Fine-tuned on MS-COCO=true2026.0645.2
READ-CLIPViT=B/32, Resolution=224, Fine-tuned on MS-COCO=true2026.0643.9
NegCLIPViT=B/32, Resolution=224, Fine-tuned on MS-COCO=true2026.0642.4
CLIPViT=L/14, Resolution=336, Data=0.4B2026.0641.8
Triplet-CLIPViT=B/32, Resolution=224, Fine-tuned on MS-COCO=true2026.0641.6
GNM-CLIPViT=B/32, Resolution=224, Fine-tuned on MS-COCO=true2026.0641.6
CLIPViT=B/32, Resolution=224, Data=0.4B2026.0641.2
CE-CLIPViT=B/32, Resolution=224, Fine-tuned on MS-COCO=true2026.0640.7
CLIPViT=L/14, Resolution=224, Data=0.4B2026.0640.7