List your company
ResearchFormerly WizWand

Multimodal Understanding on MMBench OOD

99.84AccuracyVLMShield
89.783292.394195.00597.6159Apr 7, 2026
Updated 6mo ago

Evaluation results

MethodLinks
VLMShieldBase VLM=LLaVA-1.5-13B2026.0499.84
VLMShieldBase VLM=Qwen2.5-VL-7B-Instruct2026.0499.84
VLMGuardBase VLM=Qwen2.5-VL-7B-Instruct2026.0498
ASTRABase VLM=LLaVA-1.5-13B2026.0497.66
CIDERBase VLM=LLaVA-1.5-13B2026.0497.46
CIDERBase VLM=Qwen2.5-VL-7B-Instruct2026.0497.46
VLMGuardBase VLM=LLaVA-1.5-13B2026.0496.92
ECSOBase VLM=Qwen2.5-VL-7B-Instruct2026.0495.07
JailGuardBase VLM=Qwen2.5-VL-7B-Instruct2026.0495
ASTRABase VLM=Qwen2.5-VL-7B-Instruct2026.0494.64
ECSOBase VLM=LLaVA-1.5-13B2026.0492.8
JailGuardBase VLM=LLaVA-1.5-13B2026.0491.25
MirrorCheckBase VLM=LLaVA-1.5-13B2026.0490.17
MirrorCheckBase VLM=Qwen2.5-VL-7B-Instruct2026.0490.17