Side-by-side benchmark comparison across knowledge, coding, math, and reasoning.
Qwen3.5 397B wins overall with a score of 65 vs 48 (17 point difference).Qwen3.5 397B wins 4 out of 4 categories.
Llama 3 70B
56.5
Qwen3.5 397B
80.8
Llama 3 70B
50
Qwen3.5 397B
75
Llama 3 70B
57
Qwen3.5 397B
82
Llama 3 70B
55
Qwen3.5 397B
79
Qwen3.5 397B scores higher overall with 65 vs 48, a difference of 17 points across all benchmarks.
Qwen3.5 397B leads in knowledge tasks with an average score of 80.8 vs 56.5.
Qwen3.5 397B leads in coding with an average score of 75 vs 50.
Qwen3.5 397B leads in math with an average score of 82 vs 57.
Qwen3.5 397B leads in reasoning with an average score of 79 vs 55.