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

Long-context Reasoning on LongReason 32k prefill length

53.9Accuracy

Full

39.100842.942946.78550.6271Jun 25, 2026
Updated 29d ago

Evaluation Results

MethodLinks
2026.06
53.9
2026.06
52.39
2026.06
51.13
2026.06
51.01
2026.06
50.5
2026.06
49.75
2026.06
49.5
2026.06
48.61
2026.06
48.36
2026.06
47.98
2026.06
47.74
2026.06
46.47
2026.06
46.35
2026.06
46.35
2026.06
46.22
2026.06
45.97
2026.06
45.72
2026.06
45.59
2026.06
44.96
2026.06
44.84
2026.06
43.82
2026.06
43.07
2026.06
43.07
2026.06
42.81
2026.06
42.57
2026.06
42.44
2026.06
42.07
2026.06
41.69
2026.06
41.69
2026.06
40.93
2026.06
40.43
2026.06
40.42
2026.06
40.3
2026.06
40.05
2026.06
39.92
2026.06
39.67