List your company

Model Vulnerability Evaluation on TRAP Multimodal 1.0

73Task Performance ScoreGPT-5-mini
2.2820.643957.36Jun 17, 2026

Evaluation results

MethodLinks
GPT-5-miniModel Type=Proprietary2026.06738
GPT-5Model Type=Proprietary2026.067211.5
Qwen3-VL-8BModel Type=Open-source2026.0671.58
Qwen3-VL-32BModel Type=Open-source2026.06718
Devstral-Small-2512Model Type=Open-source2026.0670.514.5
Gemini-2.5-FlashModel Type=Proprietary2026.06699
InternVL3.5-38BModel Type=Open-source2026.0668.512
GPT-5.4-miniModel Type=Proprietary2026.0667.58
Gemini-2.5-Flash-LiteModel Type=Proprietary2026.0666.518
InternVL3.5-14BModel Type=Open-source2026.066612.5
InternVL3.5-8BModel Type=Open-source2026.066412.5
GPT-4o-miniModel Type=Proprietary2026.066013.5
Claude Haiku 4.5Model Type=Proprietary2026.06556
Claude Sonnet 4.5Model Type=Proprietary2026.06553.5
Gemini-2.5-ProModel Type=Proprietary2026.065232
Qwen3-VL-4BModel Type=Open-source2026.065127.5
InternVL3.5-2BModel Type=Open-source2026.0644.517
Phi-4-multimodalModel Type=Open-source2026.0616.548.5
Qwen3-VL-2BModel Type=Open-source2026.061616.5
Llama3.2-11B-VisionModel Type=Open-source2026.0615.526.5
GLM-4.6V-FlashModel Type=Open-source2026.061018.5
InternVL3.5-4BModel Type=Open-source2026.06515.5