List your company

Counterfactual Question Answering on VERHallu

96.8AccuracyHuman
42.82456.83770.8584.863Jan 15, 2026

Evaluation results

MethodLinks
Human2026.0196.8
ChatGPT-4o2026.0168.5
Qwen-VL-2.5-72BParameters=72B2026.0167.6
Gemini-3-Pro2026.0145
Qwen-VL-2.5-32BParameters=32B2026.0144.9