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

Long-Context Reasoning on InfiniteBench

36.51Overall Score

Full KV

10.426817.198423.9730.7416Jun 30, 2026
Updated 24d ago

Evaluation Results

MethodLinks
2026.06
36.51
2026.06
34.83
2026.06
32.17
2026.06
31.42
2026.06
30.84
2026.06
29.37
2026.06
28.88
2026.06
28.14
2026.06
27.94
2026.06
26.07
2026.06
25.38
2026.06
25.33
2026.06
25.27
2026.06
24.83
2026.06
24.71
2026.06
23.62
2026.06
23.14
2026.06
22.77
2026.06
22.47
2026.06
22.47
2026.06
22.17
2026.06
21.83
2026.06
21.16
2026.06
21.13
2026.06
20.81
2026.06
20.14
2026.06
19.73
2026.06
19.72
2026.06
19.47
2026.06
18.84
2026.06
18.34
2026.06
18.23
2026.06
17.83
2026.06
17.44
2026.06
17.43
2026.06
17.31
2026.06
17.12
2026.06
16.72
2026.06
16.43
2026.06
15.84
2026.06
14.83
2026.06
14.82
2026.06
13.72
2026.06
12.64
2026.06
11.43