GLM 5.2

AI reference score 6.53 ± 0.88 · max (best) Last 4 weekly runs · 12 images
Tier comparison
TierSamplesAI reference scoreValid SVG rateMean timeSamples
max
GLM 5.2 · max · 2026-10-25 · #3 · 6.48
6.53 ± 0.88Last 4 weekly runs · 12 images 100%37 s3
high default
GLM 5.2 · high · 2026-10-25 · #1 · 5.92
5.14 ± 0.76Last 4 weekly runs · 12 images 100%16 s3

Latest samples

max

AI reference score 6.53 ± 0.88 Last 4 weekly runs · 12 images Monitoring

high

AI reference score 5.14 ± 0.76 Last 4 weekly runs · 12 images Monitoring

Score trend

max Normal

2026-10-08 · 7.752026-10-09 · 7.752026-10-10 · 7.752026-10-11 · 5.992026-10-12 · 5.992026-10-13 · 5.992026-10-14 · 5.992026-10-15 · 5.992026-10-16 · 5.992026-10-17 · 5.992026-10-18 · 6.322026-10-19 · 6.322026-10-20 · 6.322026-10-21 · 6.322026-10-23 · 6.322026-10-24 · 6.322026-10-25 · 6.062026-10-26 · 6.06

18 records in the last 30 days

high Normal

2026-10-08 · 5.142026-10-09 · 5.142026-10-10 · 5.142026-10-11 · 5.522026-10-12 · 5.522026-10-13 · 5.522026-10-14 · 5.522026-10-15 · 5.522026-10-16 · 5.522026-10-17 · 5.522026-10-18 · 4.862026-10-19 · 4.862026-10-20 · 4.862026-10-21 · 4.862026-10-23 · 4.862026-10-24 · 4.862026-10-25 · 5.032026-10-26 · 5.03

18 records in the last 30 days

Earlier samples

max

high

Versions by score

  1. GLM 5.3
    GLM 5.3 · default
    default · This model version has no selectable reasoning tiers 8.34Last 3 weekly runs · 9 images
  2. GLM 5.2
    GLM 5.2 · max
    max 6.53Last 4 weekly runs · 12 images

Scores by criterion

Scores by tier
Scoring criteria max high
Pelican likeness6.505.56
Bicycle correctness6.225.28
Riding pose5.393.72
Composition6.566.06
Obvious errors (higher is better)5.444.44

Release notes

Original (Chinese)先向全体 GLM Coding Plan 用户公开可用。官方 changelog、博客、API、Chat 与 MIT 开源权重于 2026-06-16 发布:https://docs.z.ai/release-notes/new-released 、https://www.zhipuai.cn/zh/research/161 。

· View release source ↗