Side-by-side benchmark comparison across knowledge, coding, math, and reasoning.
GLM-5 wins overall with a score of 68 vs 20 (48 point difference).GLM-5 wins 4 out of 4 categories.
GLM-5
85
Mistral 8x7B v0.2
26.8
GLM-5
80
Mistral 8x7B v0.2
21
GLM-5
87
Mistral 8x7B v0.2
28
GLM-5
83
Mistral 8x7B v0.2
26
GLM-5 scores higher overall with 68 vs 20, a difference of 48 points across all benchmarks.
GLM-5 leads in knowledge tasks with an average score of 85 vs 26.8.
GLM-5 leads in coding with an average score of 80 vs 21.
GLM-5 leads in math with an average score of 87 vs 28.
GLM-5 leads in reasoning with an average score of 83 vs 26.