List your company

Hallucination Benchmark on POPE A

87.6Accuracy (POPE A)V-STAR
78.7681.05583.3585.645Apr 11, 2026

Evaluation results

MethodLinks
V-STARData Size=40k2026.0487.6
ThinkLite-VLData Size=11k2026.0487.1
VL-CogitoData Size=80k2026.0484.1
Vision-R1Data Size=210k2026.0484
Qwen2.5VL2026.0483.1
VL-RethinkerData Size=39k2026.0482.8
R1-OnevisionData Size=155k2026.0482.5
OpenVLThinkerData Size=59.2k2026.0479.1