List your company

Cross-modal Hallucination Detection on Curse of Multi-Modalities (CMM) 1.0 (test)

96.5VL Precision (pa)Qwen 3 Omni
74.1479.94585.7591.555Mar 3, 2026

Evaluation results

MethodLinks
Qwen 3 Omni2026.0396.577.5938897.59495.56094.56793659575.3
VITA 1.52026.0392.591.542.530.577.772.95724.477.467.78655.572.257.1
Qwen 2.5 Omni + MoD-DPO++base_model=Qwen 2.5 Omni, alignment_method=MoD-DPO++2026.0392.598.592.785.894.595.591.375.189.184.374.983.889.287.2
Qwen 2.5 Omni + MoD-DPObase_model=Qwen 2.5 Omni, alignment_method=MoD-DPO2026.0391.598.592.383.294.594.592.374.288.483.873.983.188.886.2
MiniCPM-O 2.6 + MoD-DPObase_model=MiniCPM-O 2.6, alignment_method=MoD-DPO2026.0390.595.5927691.58682.584.88978.982.473.78882.5
MiniCPM-O 2.6 + MoD-DPO++base_model=MiniCPM-O 2.6, alignment_method=MoD-DPO++2026.039096.592.577.591.58783.585.688.479.683.875.188.383.6
Qwen 2.5 Omni + OmniDPObase_model=Qwen 2.5 Omni, alignment_method=OmniDPO2026.03899889.583.5949388.869.986.181.372.282.686.684.7
OmniVinci2026.0388.58174.59098729163.59635877589.269.4
Qwen 2.5 Omni2026.03889889.984.49393.587.867.785.380.774.283.286.484.6
MiniCPM-O 2.6 + OmniDPObase_model=MiniCPM-O 2.6, alignment_method=OmniDPO2026.0388959176.588.58479.178.888.477.883.271.686.480.6
MiniCPM-O 2.62026.03879490.574878479.580.587.476.48273.785.680.4
MiniCPM-O 2.6 + DPObase_model=MiniCPM-O 2.6, alignment_method=DPO2026.03879389.574.58684.579.579.587.87681.574.385.280.3
Qwen 2.5 Omni + DPObase_model=Qwen 2.5 Omni, alignment_method=DPO2026.038698.589.586.592.494.486.763.584.380.472.584.185.284.6
VideoLLaMA 22026.03758677.59478986275.5809057.54371.781.1