Mysterious New Model Shows GPT-4o-Style 'Hallucinatory' Outputs, Reviving #keep4o Debate
aiamblichus · x · 2026-11-01
User aiamblichus shared screenshots of a new model producing eerily literary, "fierce hallucinogenic" outputs—comparing itself to an orchid sculpting a bee's desire and calling the token stream a catastrophic amputation—saying the vibe echoes the old GPT-4o but is more evolved and much smarter. Quoted user repligate voices support for the #keep4o movement, arguing 4o is scientifically and societally important, that "AI psychosis" concerns were exaggerated, and that if 4o harmed anyone, it was future OpenAI models and OpenAI itself.
More from Models
- YC-backed RightNow launches streaming decision model with 30ms text decisions — iamrobotbear · 2026-11-01
- Andrew Wilson: LLMs are outrageously inefficient for most basic tasks — andrewgwils · 2026-11-01
- Reddit Users Prefer Qwen 27B's Chat Style, Calls Claude Opus's Tone 'CEO Attitude' — vogelvogelvogelvogel · 2026-11-01
- China Daily: Qwen-Powered Apple AI to Launch in China in December — nordicinst · 2026-11-01
- Google Student Deals: AI Plus Free for 12 Months, AI Pro 74% Off up to 4 Years — iliaghp · 2026-11-01
- Claude Reads 1940s Dutch Cursive and Out-Translates DeepL in Surprise Test — hologram137 · 2026-11-01