List your company
ResearchFormerly WizWand

Multimodal Hallucination Evaluation on HallBench

65.2ScoreGPT-5-Thinking
48.35252.72657.161.474Sep 30, 2025
Updated 7mo ago

Evaluation results

MethodLinks
GPT-5-ThinkingModel Category=Close-source Models2025.0965.2
Gemini-2.5-ProModel Category=Close-source Models2025.0964.1
VAPO-Thinker-7BModel Category=Our Models2025.0957.4
VLAA-Thinker-7BModel Category=Open-source Models2025.0954.7
R1-OneVision-7BModel Category=Open-source Models2025.0952.5
Qwen2.5-VL-7BModel Category=Open-source Models2025.0950
Vision-R1-7BModel Category=Open-source Models2025.0949.5
VAPO-Thinker-3BModel Category=Our Models2025.0949.5
InternVL2.5-8BModel Category=Open-source Models2025.0949