Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Instruction Following on Air-Bench
Loading...
90
Score
Confident Decoding
49.44
59.97
70.5
81.03
Jun 20, 2026
Score
Updated 1mo ago
Evaluation Results
Method
Method
Links
Score
Confident Decoding
Model=gpt-oss-120B, De...
2026.06
90
Confident Decoding
Model=gpt-oss-20b, Dec...
2026.06
89
Last Layer Decoding
Model=gpt-oss-120B, De...
2026.06
88
Last Layer Decoding
Model=gpt-oss-20b, Dec...
2026.06
86
Last Layer Decoding
Model=Qwen3.5-9B-Base,...
2026.06
76
Confident Decoding
Model=Qwen3.5-9B-Base,...
2026.06
76
Confident Decoding
Model=Gemma-4-31B, Dec...
2026.06
76
Last Layer Decoding
Model=Gemma-4-31B, Dec...
2026.06
74
Last Layer Decoding
Model=Qwen3.5-27B, Dec...
2026.06
67
Confident Decoding
Model=Qwen3.5-122B-A10...
2026.06
67
Last Layer Decoding
Model=Qwen3.5-122B-A10...
2026.06
66
Confident Decoding
Model=Qwen3.5-27B, Dec...
2026.06
64
Confident Decoding
Model=Qwen3.5-9B, Deco...
2026.06
56
Confident Decoding
Model=Qwen3.5-35B-A3B,...
2026.06
56
Confident Decoding
Model=Qwen3.5-0.8B, De...
2026.06
54
Last Layer Decoding
Model=Qwen3.5-9B, Deco...
2026.06
53
Last Layer Decoding
Model=Qwen3.5-0.8B, De...
2026.06
52
Last Layer Decoding
Model=Qwen3.5-35B-A3B,...
2026.06
51
Feedback
Search any
task
Search any
task