Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

General VQA on HallusionBench

73.48Accuracy

Gemini 3-Pro

40.813649.294357.77566.2557Feb 4, 2026Feb 11, 2026Feb 18, 2026Feb 25, 2026Mar 4, 2026Mar 11, 2026Mar 18, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
73.48
2026.02
66.58
2026.02
64.01
2026.02
63.87
63.7
2026.03
51.89
2026.03
48.18
2026.03
46.54
2026.03
42.07