LLMs Lack Intuition for Thinking Time

goodside · x · 2026-07-14

LLMs seem to lack a stable intuition for "how long their own thinking takes".

goodside points out that ChatGPT 5.6 Pro displays response times, making this phenomenon easier to observe; he added that Claude Fable 5 Max has a similar issue, spending less than 2 minutes thinking on a prompt, yet later "knowing" it didn't actually think for an hour.

Related event: LLMs Lack Accurate Intuition About Their Own Thinking Time(2 posts)→

Original post →

More from Models

Models channel →