Anshu(@anshuc):GLM-5.3 is really fun for web design. I redid my old GLM-5.2 demo with 5.3 and WOW, look at the difference! Zero dependencies, zero assets. Just an HTML file. It one-shot most of this, but I had to a few iterations after to fix visual glitches and bugs. It does need a little handholding. Not quite on par with Claude Opus/Fable, but not that far off either. The big drawback is it can't see images. It made a valiant effort to find and diagnose issues by writing scripts to analyze pixel values of screenshots. They really need to fix this. In the meantime, you could probably let it use some small model as its eyes to work more autonomously. I'll have to try that next. Impressive model, congrats @Zai

2026.08.19 20:24

GLM-5.3 is really fun for web design. I redid my old GLM-5.2 demo with 5.3 and WOW, look at the difference! Zero dependencies, zero assets. Just an HTML file. It one-shot most of this, but I had to a few iterations after to fix visual glitches and bugs. It does need a little handholding. Not quite on par with Claude Opus/Fable, but not that far off either. The big drawback is it can't see images. It made a valiant effort to find and diagnose issues by writing scripts to analyze pixel values of screenshots. They really need to fix this. In the meantime, you could probably let it use some small model as its eyes to work more autonomously. I'll have to try that next. Impressive model, congrats @Zai_org!

显示更多

Anshu@anshuc

2026.06.17 00:00

Whoa, GLM-5.2 is INSANE for UI/UX with the right prompting. Open models have finally closed the gap. You have to push it, but it's super creative. I just asked for a beautiful personal site; this was all the model's idea (multiple turns):

显示更多