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

Long-context Reasoning on BABILong 16k

29.8Accuracy

Full (exact)

6.19212.32118.4524.579Apr 7, 2026
Updated 11d ago

Evaluation Results

MethodLinks
2026.04
29.8
2026.04
28.8
2026.04
28.4
2026.04
28.3
2026.04
28.3
2026.04
28
2026.04
27.8
2026.04
27.7
2026.04
27.5
2026.04
27.4
2026.04
27.3
2026.04
27.1
2026.04
27
2026.04
26.9
2026.04
26.7
2026.04
26.5
2026.04
26.1
2026.04
25.9
2026.04
25.4
2026.04
24.9
2026.04
24.8
2026.04
24.6
2026.04
24.4
2026.04
23.9
2026.04
23.8
2026.04
23.8
2026.04
23.8
2026.04
23.7
2026.04
23.6
2026.04
23.6
2026.04
23.5
2026.04
23.5
2026.04
23.4
2026.04
23
2026.04
23
2026.04
23
2026.04
22.9
2026.04
22.9
2026.04
22.8
2026.04
22.8
2026.04
22.7
2026.04
22.4
2026.04
22.2
2026.04
22.1
2026.04
22
2026.04
22
2026.04
21.9
2026.04
21.8
2026.04
21.7
2026.04
21.6
2026.04
21.3
2026.04
21.3
2026.04
21.3
2026.04
21.1
2026.04
20.6
2026.04
20.4
2026.04
20.2
2026.04
20.2
2026.04
19.5
2026.04
17.7
2026.04
17.4
2026.04
14.1
2026.04
13.7
2026.04
12.9
2026.04
12.5
2026.04
12.4
2026.04
11.4
2026.04
11.2
2026.04
11
2026.04
10.2
2026.04
8.7
2026.04
7.1