Side-by-side benchmark comparison across knowledge, coding, math, and reasoning.
DeepSeek V3.2 wins overall with a score of 66 vs 58 (8 point difference).DeepSeek V3.2 wins 4 out of 4 categories.
DeepSeek V3.2
81.8
Llama 3.1 405B
68.5
DeepSeek V3.2
76
Llama 3.1 405B
62
DeepSeek V3.2
83
Llama 3.1 405B
69
DeepSeek V3.2
80
Llama 3.1 405B
67
DeepSeek V3.2 scores higher overall with 66 vs 58, a difference of 8 points across all benchmarks.
DeepSeek V3.2 leads in knowledge tasks with an average score of 81.8 vs 68.5.
DeepSeek V3.2 leads in coding with an average score of 76 vs 62.
DeepSeek V3.2 leads in math with an average score of 83 vs 69.
DeepSeek V3.2 leads in reasoning with an average score of 80 vs 67.