Side-by-side benchmark comparison across knowledge, coding, math, and reasoning.
GLM-4.5-Air wins overall with a score of 26 vs 21 (5 point difference).GLM-4.5-Air wins 4 out of 4 categories.
GLM-4.5-Air
32.8
Mistral 7B v0.3
27.8
GLM-4.5-Air
27
Mistral 7B v0.3
22
GLM-4.5-Air
34
Mistral 7B v0.3
29
GLM-4.5-Air
32
Mistral 7B v0.3
27
GLM-4.5-Air scores higher overall with 26 vs 21, a difference of 5 points across all benchmarks.
GLM-4.5-Air leads in knowledge tasks with an average score of 32.8 vs 27.8.
GLM-4.5-Air leads in coding with an average score of 27 vs 22.
GLM-4.5-Air leads in math with an average score of 34 vs 29.
GLM-4.5-Air leads in reasoning with an average score of 32 vs 27.