GLM

62.9 Worse than usual experience score ▼ 2.8 Confidence: Medium

240 opinions tracked this week · 7-day rolling window

30-day experience score

Version breakdown

Only comments that explicitly name a version contribute to its score.

GLM 5.2	61.5−1.4	n=170	Confidence: Medium
GLM 5.1	65.3+2.4	n=7	Confidence: Low
GLM 4.7	mentions only	n=3	Confidence: Insufficient
GLM 4.5 Air	mentions only	n=2	Confidence: Insufficient
GLM 4.7 Flash	mentions only	n=1	Confidence: Insufficient

55 generic family mentions are included in the family score, not any version score.

By experience dimension

General text	59	n=79
Coding	71	n=38
Reasoning	76	n=48
Image / vision	not enough data	n=1
Roleplay / creative	not enough data	n=2
Safety & refusals	not enough data	n=4
Speed & latency	48	n=58
Local deploy	46	n=10

Top positive drivers

strong reasoning Reasoning ×31
speed quality Speed & latency ×19
competitor flight General text ×15
problem solved General text ×14
competitor flight Coding ×7

Top negative drivers

performance issue Speed & latency ×12
speed quality Speed & latency ×9
competitor flight General text ×9
product limit General text ×5
pricing billing General text ×4

What users are saying

🟢 “还有glm从年初开始就是断档的国模coding第一，人家资历还老，怎么说也是glm最先最强”
via zhihu · coding · zh

🟢 “智谱专注于Coding是无比正确的选择，说明他们在RL上探出路子了，国内唯一”
via zhihu · reasoning · zh

🟢 “That's surprisingly good imo. My M3 Ultra gets ~15tps”
via reddit · speed_latency · en

🔴 “Hermes does that , Cline does that , Opencode does that , nothing spectacular. its all agent skills.”
via reddit · text_general · en