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

Long-context Reasoning on LongReason 64k prefill length

53.02Accuracy

Full

36.650440.900245.1549.3998Jun 25, 2026
Updated 29d ago

Evaluation Results

MethodLinks
2026.06
53.02
2026.06
49.87
2026.06
48.99
2026.06
48.74
2026.06
48.61
2026.06
48.11
2026.06
47.73
2026.06
47.61
2026.06
47.48
2026.06
46.47
2026.06
46.22
2026.06
45.72
2026.06
45.72
2026.06
45.59
2026.06
45.34
2026.06
45.04
2026.06
44.96
2026.06
44.28
2026.06
43.53
2026.06
42.7
2026.06
41.56
2026.06
41.31
2026.06
41.18
2026.06
41.06
2026.06
40.93
2026.06
40.81
2026.06
39.42
2026.06
39.29
2026.06
39.17
2026.06
38.79
2026.06
38.53
2026.06
38.41
2026.06
38.28
2026.06
38.04
2026.06
37.66
2026.06
37.28