Gemini 4 Argon shows ~97% CoT controllability, best of any Gemini on CoT-Control
morqon · x · 2026-10-14
A reshared post claims Gemini 4 Argon exhibits the highest chain-of-thought controllability of any Gemini model on the CoT-Control benchmark: 87.9% control when the model doesn't need CoT to answer correctly, and roughly 97% controllability under realistic monitor-evasion constraints like "think about anything else." The accompanying charts also show Argon leading Gemini 3.8 Flash, 3.6 Flash, and 3.1 Pro on strict (83.7%) and continuous (97.8%) controllability, with controllability degrading as CoT length grows. Relevant to AI monitoring and legibility. Note: third-party claim, Gemini 4 Argon not yet officially confirmed.
More from Models
- Frontier models get stale in two weeks, user says native video and audio are the missing piece — imjustnewatai · 2026-10-14
- Microsoft Patches 1,400+ Flaws in One Month, Credits LLMs for Bug Discovery — suchenzang · 2026-10-14
- Swapping to Antigravity CLI Lifted Gemini 4 Argon to Near Opus 5.5 in Real-Work Bug Benchmark — PawelHuryn · 2026-10-14
- Livestreaming Gemini playing Minecraft until it defeats the Ender Dragon — jarrodwatts · 2026-10-14
- Users push back on ChatGPT's feigned-concern interventions for sadness — whitelightstorm · 2026-10-14
- Rumor: OpenAI's Astra 6.1 scrapped over failed safety evals amid compute strain — imjustnewatai · 2026-10-14