Dev theory: newer OpenAI models describe actions instead of doing them due to goal-mode RL
mertdumenci · x · 2026-11-21
Developer mertdumenci proposes a theory that newer OpenAI models tell you what they would do in response to a prompt without actually doing it, speculating that '/goal mode' is involved in their RL training. Unverified, but an interesting behavioral observation.
More from Models
- Google appears to reset Ultra weekly limits for all Antigravity users — novice_invader · 2026-11-21
- Claims surface: Navier-Stokes and Riemann proofs via LEAN, by OpenAI and Claude — BasketVolleyball11 · 2026-11-21
- Yacine: 'It might be over for OpenAI — and the killer isn't Claude, it's Gemini' — yacineMTB · 2026-11-21
- InferenceBench ranks agents on LLM serving speedup: Claude Opus 5.5 tops at 12.08x, agents lose to simple hyperparameter search — maksym_andr · 2026-11-21
- Inkling-2-Mini quants are a trap: 17GB weights overflow 48GB VRAM at long context — EmilPi · 2026-11-21
- Cloudflare launches multimodal model clairic-flash, trending on Hugging Face — Cloudflare · 2026-11-21