Dev slams Claude's safety refusals in security work, argues GLM 5.3 is a frontier model
gnukeith · x · 2026-10-25
Developer gnukeith weighed in on the "GLM vs Claude" debate, arguing there's no clear winner — it depends on the task. He says Claude can find only minor vulnerabilities in his codebase, then screams at him because it thinks he's hacking the government, and predicts OpenAI models will do the same by requiring KYC to access security features. He maintains GLM 5.3 is a frontier model and neither clearly beats the other at coding.
More from Models
- Hugging Face researcher early impressions: Gemini 4 output "really nice" — eliebakouch · 2026-10-25
- 2026 hallucination leaderboard: 141 of 202 models err more than they answer — gerardsans · 2026-10-25
- Google Translate handles song lyrics but Gemini refuses to, user notes — adrianscottcom · 2026-10-25
- Reddit user: Kimi K3 Pro shows little edge over DeepSeek 4.1 Flash — Hannibalj2ca · 2026-10-25
- Matthew Berman claims Kimi K3.1 invented its benchmark scores — Matthew Berman · 2026-10-25
- Figure built a secret HQ lab called Blackbox; Adcock calls new model "baby AGI" — adcock_brett · 2026-10-25