Side-by-side benchmark comparison across knowledge, coding, math, and reasoning.
Mistral 8x7B wins overall with a score of 52 vs 22 (30 point difference).Mistral 8x7B wins 4 out of 4 categories.
GPT-OSS 20B
28.8
Mistral 8x7B
62.8
GPT-OSS 20B
23
Mistral 8x7B
55
GPT-OSS 20B
30
Mistral 8x7B
64
GPT-OSS 20B
28
Mistral 8x7B
62
Mistral 8x7B scores higher overall with 52 vs 22, a difference of 30 points across all benchmarks.
Mistral 8x7B leads in knowledge tasks with an average score of 62.8 vs 28.8.
Mistral 8x7B leads in coding with an average score of 55 vs 23.
Mistral 8x7B leads in math with an average score of 64 vs 30.
Mistral 8x7B leads in reasoning with an average score of 62 vs 28.