‹ BackHN Continuity

Thread

GPT-6 Sol and Luna

1779 points · 855 comments · OfficialTurkey

  1. hamburglar1 · · focus · HN ↗
    Code deception 10% at 5.6 to 1.3% for 6.0? So models are getting more safe rather than less safe? hmmm
Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.