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

Long-context evaluation on RULER 4k

93.73Score

QUOKA

16.270836.380456.4976.5996Feb 9, 2026
Updated 22d ago

Evaluation Results

MethodLinks
2026.02
93.73-
2026.02
89.97-
2026.02
87.93-
2026.02
87.85-
2026.02
87.84-
2026.02
86.71-
2026.02
85.72-
2026.02
84.52-
2026.02
82.83-
2026.02
82.45-
2026.02
79.67-
2026.02
79.36-
2026.02
78.92-
2026.02
78.25-
2026.02
78.07-
2026.02
77.17-
2026.02
76.2-
2026.02
75.48-
2026.02
75.15-
2026.02
74.84-
2026.02
74.4-
2026.02
70.07-
2026.02
69.58-
2026.02
67.35-
2026.02
65.55-
2026.02
62.85-
2026.02
58.18-
2026.02
53.34-
2026.02
43.72-
2026.02
37.29-
2026.02
36.66-
2026.02
34.09-
2026.02
29.15-
2026.02
27.29-
2026.02
19.25-
2026.07
-77.7
2026.07
-79.2
2026.07
-69.2
2026.07
-80