We kept running the /design command against every model we could get our hands on, mostly to settle arguments about which one actually has good taste. This repo is the receipts: 20 side-by-side runs across frontier models (Opus 4.8, Sonnet 5, GPT-5.5, Fable 5) and open models (GLM 5.2, DeepSeek V4 Pro, Kimi K2.7 Code).
📖 Docs: commandcode.ai/docs/slash-commands/design
The short version: GLM 5.2 keeps matching or beating closed models on design precision, motion, and UX, for a fraction of the cost. DeepSeek V4 Pro is closing the gap fast too.
- Interactive / Game UX Tests
- Landing Pages & SaaS
- Dashboards, Portfolios & Other
- Developer Taste & Precision Tests
- How to Use These
- GLM 5.2 vs Sonnet 5
- GLM 5.2 vs Opus 4.8
- GLM 5.2 vs Sonnet 5
- GLM 5.2 vs Sonnet 5
- GLM 5.2 vs Opus 4.8
- GLM 5.2 vs Opus 4.8
- GLM 5.2 vs Opus 4.8
- Opus 4.8 vs GLM 5.2 vs GPT-5.5
- GLM 5.2 vs Opus 4.8
- GLM 5.2 vs Opus 4.8
- GLM 5.2 Fast vs DeepSeek V4 Pro vs Kimi K2.7 Code
- GLM 5.2 Fast vs GLM 5.2
- Open models
Want to run your own comparisons? Use /design, it ships with 17 modes, including redesign and checkup and many more.
Full command reference: commandcode.ai/docs/slash-commands/design
Built with Command Code.