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

Long-context Reasoning on LongBench

73.8Score

GHG-TDA

29.70441.15252.664.048Dec 12, 2025Dec 22, 2025Jan 1, 2026Jan 11, 2026Jan 21, 2026Jan 31, 2026Feb 10, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.02
73.8
2026.02
73.5
2026.02
72.6
2026.02
72.2
2026.02
69.9
2026.02
69.5
2026.02
68.7
2026.02
68.6
2026.02
68.5
2026.02
68.1
2026.02
67
2026.02
66.8
2026.02
66.4
2026.02
66
2026.02
64.64
2026.02
63.71
2026.02
63.4
2026.02
63.2
2026.02
63.1
2026.02
62.8
2026.02
62.8
2026.02
62.8
2026.02
62.4
2026.02
62
2026.02
61.7
2026.02
61.4
2026.02
61.2
2026.02
61
2026.02
60.4
2026.02
59.3
2026.02
59.3
2026.02
59.1
2026.02
59
2026.02
58.9
2026.02
58.68
2026.02
58.6
2026.02
58.4
2026.02
58.4
2026.02
58.4
2026.02
58.2
2026.02
58.1
2026.02
58
2026.02
57.7
2026.02
57.6
2026.02
57.3
2026.02
53.57
2026.02
53.4
2026.02
52.9
2026.02
52.7
2026.02
52.7
2025.12
35.1
2025.12
34.4
2025.12
33.6
2025.12
33.6
2025.12
33.3
2025.12
33.1
2025.12
32.4
2025.12
31.8
2025.12
31.8
2025.12
31.8
2025.12
31.4
2025.12
31.4