Re:November 2026 Claude Fable 5.2 Scores 65 on Artificial An...
by jrs100000 on Monday November 09, 2026 @08:56PM
(#50435104)
Has anyone tried to actually use this? Six turns, about 10 minutes total thinking, and its going in circles spitting out the sort of shallow logic I would expect from basic Deepseek. This is genuinely awful. Im planning to turn in my free reset and go back to 5.5, except that also started fucking up basic instructions a few days ago. Everyone keeps going on about these BS rankings that every single release of every model claims to ace, when we all should know have almost nothing to do with real results. I feel like Im taking crazy pills here.
Maybe its different for vibe coding, I wouldnt know.
Maybe its different for vibe coding, I wouldnt know.