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

Multiple-choice Question Answering on GPQA (accuracy)

95.2Accuracy (%)

Pass@100

-3.80821.89647.673.304Mar 13, 2026Mar 19, 2026Mar 26, 2026Apr 1, 2026Apr 8, 2026Apr 14, 2026Apr 21, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.04
95.2
2026.04
81
2026.04
73.74
2026.04
73.23
2026.04
73.23
2026.04
72.9
2026.04
70.87
2026.04
69.3
2026.04
66.8
2026.04
66.4
2026.04
60.5
2026.04
60.1
2026.04
59.1
2026.04
56.97
2026.04
56.67
2026.04
54.85
2026.04
54.85
2026.04
54.55
2026.04
47.4
2026.04
47.1
2026.04
42.9
2026.04
41.9
2026.04
40.5
2026.04
37.8
2026.04
37.6
2026.04
35.9
2026.04
30.5
2026.03
30.35
2026.03
30.16
2026.04
28.3
2026.03
27.18
2026.03
26.35
2026.03
26.32
2026.03
26.09
2026.03
25.93
2026.03
23.91
2026.03
19.87
2026.03
19.47
2026.03
18.24
2026.03
17.52
2026.03
12.51
2026.03
8.64
2026.03
8.15
2026.04
0