Side-by-side benchmark comparison across knowledge, coding, math, and reasoning.
DeepSeek-R1 wins overall with a score of 35 vs 20 (15 point difference).DeepSeek-R1 wins 4 out of 4 categories.
DeepSeek-R1
41.8
Mistral 8x7B v0.2
26.8
DeepSeek-R1
36
Mistral 8x7B v0.2
21
DeepSeek-R1
43
Mistral 8x7B v0.2
28
DeepSeek-R1
41
Mistral 8x7B v0.2
26
DeepSeek-R1 scores higher overall with 35 vs 20, a difference of 15 points across all benchmarks.
DeepSeek-R1 leads in knowledge tasks with an average score of 41.8 vs 26.8.
DeepSeek-R1 leads in coding with an average score of 36 vs 21.
DeepSeek-R1 leads in math with an average score of 43 vs 28.
DeepSeek-R1 leads in reasoning with an average score of 41 vs 26.