Side-by-side benchmark comparison across knowledge, coding, math, and reasoning.
GLM-4.5 wins overall with a score of 28 vs 20 (8 point difference).GLM-4.5 wins 4 out of 4 categories.
GLM-4.5
34.8
Mistral 8x7B v0.2
26.8
GLM-4.5
29
Mistral 8x7B v0.2
21
GLM-4.5
36
Mistral 8x7B v0.2
28
GLM-4.5
34
Mistral 8x7B v0.2
26
GLM-4.5 scores higher overall with 28 vs 20, a difference of 8 points across all benchmarks.
GLM-4.5 leads in knowledge tasks with an average score of 34.8 vs 26.8.
GLM-4.5 leads in coding with an average score of 29 vs 21.
GLM-4.5 leads in math with an average score of 36 vs 28.
GLM-4.5 leads in reasoning with an average score of 34 vs 26.