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

Long-context Reasoning on BABILong 4k

38.5Accuracy (BABILong 4k)

Full (exact)

9.69217.17124.6532.129Apr 7, 2026
Updated 10d ago

Evaluation Results

MethodLinks
2026.04
38.5
2026.04
37
2026.04
36.7
2026.04
36
2026.04
35.6
2026.04
34.3
2026.04
33.9
2026.04
33.9
2026.04
33.8
2026.04
33.7
2026.04
33.2
2026.04
33.1
2026.04
32.6
2026.04
32.3
2026.04
32.3
2026.04
32.2
2026.04
32.1
2026.04
32
2026.04
32
2026.04
31.7
2026.04
31.6
2026.04
31.5
2026.04
31.5
2026.04
31.1
2026.04
31
2026.04
31
2026.04
30.8
2026.04
30.6
2026.04
29.7
2026.04
29.7
2026.04
29.2
2026.04
29.1
2026.04
28.7
2026.04
28
2026.04
27
2026.04
26.7
2026.04
25.8
2026.04
25.8
2026.04
25.8
2026.04
25.6
2026.04
23
2026.04
22.6
2026.04
22.6
2026.04
20.8
2026.04
16.5
2026.04
15.6
2026.04
15.1
2026.04
13.6
2026.04
12.7
2026.04
11.4
2026.04
10.8