keltan comments on Recent AI model progress feels mostly like bullshit

[keltan]

I’d be interested to know if your lived experience has changed since Opus 4.5 was released?

[lc]

Completely different. The software vulnerabilities we’re finding with the most recent codex models are crazy.