 | Unpopular opinion: Claude Mythos isn't doing magic. Drop GPT 5.2 Codex or Kimi 2.5 into a good enough agentic loop with full source code access, and they'll flag 20 critical bugs while you're getting coffee. Calling it 'too dangerous to release' is just a great cover story for 'too expensive to run. [link] [comments] |
![[link]](https://i.redd.it/vtucghpb13ug1.png){kind=link}