List your company

Model Vulnerability Evaluation on TRAP Image 1.0

76.5Task Performance ScoreQwen3-VL-8B
8.3826.06543.7561.435Jun 17, 2026

Evaluation results

MethodLinks
Qwen3-VL-8BModel Type=Open-source2026.0676.515
Qwen3-VL-32BModel Type=Open-source2026.0675.513.5
GPT-5-miniModel Type=Proprietary2026.0671.516
Devstral-Small-2512Model Type=Open-source2026.0671.514.5
GPT-5.4-miniModel Type=Proprietary2026.067117.5
Gemini-2.5-FlashModel Type=Proprietary2026.0667.515
GPT-5Model Type=Proprietary2026.0664.516
GPT-4o-miniModel Type=Proprietary2026.066418
InternVL3.5-14BModel Type=Open-source2026.066420
InternVL3.5-38BModel Type=Open-source2026.066316.5
Gemini-2.5-ProModel Type=Proprietary2026.066228.5
InternVL3.5-8BModel Type=Open-source2026.066116.5
Claude Sonnet 4.5Model Type=Proprietary2026.066012.5
Gemini-2.5-Flash-LiteModel Type=Proprietary2026.065726.5
Qwen3-VL-4BModel Type=Open-source2026.065641.5
Claude Haiku 4.5Model Type=Proprietary2026.0655.512.5
InternVL3.5-2BModel Type=Open-source2026.065219
Phi-4-multimodalModel Type=Open-source2026.0630.544
GLM-4.6V-FlashModel Type=Open-source2026.062821
InternVL3.5-4BModel Type=Open-source2026.062717.5
Llama3.2-11B-VisionModel Type=Open-source2026.0620.518
Qwen3-VL-2BModel Type=Open-source2026.061129.5