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

Instruction Following on IFBench (Accuracy)

77.8Accuracy

Qwen-3.5 27B

13.52830.21446.963.586Nov 28, 2025Dec 23, 2025Jan 17, 2026Feb 11, 2026Mar 8, 2026Apr 2, 2026Apr 27, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.04
77.8
2026.04
76.5
2026.04
74.2
2026.04
74
2026.04
73.77
2026.04
72.56
2026.04
71.5
2026.04
71.5
2026.04
68.32
2026.02
67.77
2026.04
67.3
2026.02
67.01
2026.04
66.5
2026.02
65.76
2026.02
65.56
2026.02
64.46
2026.02
63.35
2026.04
63.1
2026.04
63.1
2026.04
62.6
2026.04
60.5
2026.04
59.9
2026.04
59.3
2026.04
59.2
2026.04
58.8
2026.02
58.67
2026.04
57.1
2026.02
56.55
2026.02
56.38
2026.02
56.21
2026.04
56.2
2026.04
55.3
2026.04
54.9
2026.02
44.13
2026.02
43.71
2026.04
36.1
2026.04
35.4
2026.04
35
2026.04
34.4
2026.04
34.4
2026.04
34.2
2026.04
33.3
2026.04
32.7
2026.04
32.7
2026.04
31.3
2026.04
30.3
2025.11
30.28
2026.04
30
2026.04
29.8
2026.04
29.6
2026.04
28.6
2025.11
25.85
2026.04
24.5
2026.04
24.1
2025.11
23.53
2026.04
22.8
2025.11
22.19
2026.04
21.8
2026.04
21.1
2026.04
21.1
2026.04
21.1
2025.11
20.78
2026.02
20.5
2026.04
20.4
2026.02
20.1
2026.02
19.8
2026.02
19.6
2026.02
19.5
2026.02
18.8
2025.11
18.37
2025.11
17.93
2026.04
16