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

Mathematical & Scientific Reasoning on GPQA

68.54Accuracy

GRPO

2.2419.452536.66553.8775Dec 15, 2025Jan 8, 2026Feb 2, 2026Feb 27, 2026Mar 24, 2026Apr 18, 2026May 13, 2026
Updated 20d ago

Evaluation Results

MethodLinks
2026.05
68.54
2026.05
67.79
2026.05
66.36
2025.12
39.06
2025.12
35.49
2025.12
32.37
2025.12
31.25
2025.12
31.25
2025.12
30.36
2025.12
30.13
2025.12
29.46
2025.12
28.35
2026.01
18.43
2026.01
17.86
2026.01
17.8
2026.01
15.66
2026.01
15.48
2026.01
7.07
2026.01
6.7
2026.01
5.27
2026.01
5.18
2026.01
4.79