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

Long-context retrieval and aggregation on RULER 32k (Average Accuracy)

89.56Average Accuracy

Full KV

17.831236.453155.07573.6969May 26, 2026
Updated 7d ago

Evaluation Results

MethodLinks
2026.05
89.56
2026.05
88.75
2026.05
88.15
2026.05
85.78
2026.05
85.41
2026.05
85.25
2026.05
85.01
2026.05
82.61
2026.05
82.58
2026.05
82.52
2026.05
78.11
2026.05
77.26
2026.05
76.17
2026.05
75.54
2026.05
75.4
2026.05
75.32
2026.05
75
2026.05
74.88
2026.05
74.85
2026.05
73.85
2026.05
73.11
2026.05
71.91
2026.05
71.28
2026.05
70.89
2026.05
70.32
2026.05
69.15
2026.05
68.62
2026.05
68.36
2026.05
66.98
2026.05
66.9
2026.05
66.04
2026.05
64.41
2026.05
64.4
2026.05
64.2
2026.05
63.68
2026.05
63.38
2026.05
62.6
2026.05
62.55
2026.05
62.42
2026.05
62.33
2026.05
61.64
2026.05
60.29
2026.05
60.27
2026.05
58.98
2026.05
58.93
2026.05
57.06
2026.05
55.77
2026.05
54.63
2026.05
53.35
2026.05
50.97
2026.05
50.93
2026.05
50.5
2026.05
50.47
2026.05
50.25
2026.05
50.23
2026.05
49.88
2026.05
47.96
2026.05
47.87
2026.05
47.21
2026.05
47.14
2026.05
43.4
2026.05
43.22
2026.05
42.76
2026.05
42.17
2026.05
38.18
2026.05
36.82
2026.05
36.09
2026.05
35.61
2026.05
34.81
2026.05
31.7
2026.05
31.65
2026.05
30.63
2026.05
29.17
2026.05
27.99
2026.05
25.7
2026.05
20.59