Side-by-side benchmark comparison across knowledge, coding, math, and reasoning.
GLM-4.5 wins overall with a score of 28 vs 21 (7 point difference).GLM-4.5 wins 4 out of 4 categories.
GLM-4.5
34.8
Mistral 7B v0.3
27.8
GLM-4.5
29
Mistral 7B v0.3
22
GLM-4.5
36
Mistral 7B v0.3
29
GLM-4.5
34
Mistral 7B v0.3
27
GLM-4.5 scores higher overall with 28 vs 21, a difference of 7 points across all benchmarks.
GLM-4.5 leads in knowledge tasks with an average score of 34.8 vs 27.8.
GLM-4.5 leads in coding with an average score of 29 vs 22.
GLM-4.5 leads in math with an average score of 36 vs 29.
GLM-4.5 leads in reasoning with an average score of 34 vs 27.