Joe Hansen(@joehansen):Artificial Analysis just launched a personality benchmark for AI models. 36 models, 900 questions. Grok 4.7 is the only model in the top 3 on both: - Conviction (direct, confident, willing to push back): #3 - Least slop (filler and generic phrasing): #3 Claude Fable is more direct but pads its answers more. GPT-6 Sol has less filler but is far less direct. That is the mix I want from an assistant. Give me the answer and stop.

Joe Hansen

@joehansen

加入 July 2014

4.4K 正在關注 22.8K 粉絲

Joe Hansen@joehansen

2026.10.19 20:05

Artificial Analysis just launched a personality benchmark for AI models. 36 models, 900 questions. Grok 4.7 is the only model in the top 3 on both: - Conviction (direct, confident, willing to push back): #3 -# Least slop (filler and generic phrasing): #3# Claude Fable is more direct but pads its answers more. GPT-6 Sol has less filler but is far less direct. That is the mix I want from an assistant. Give me the answer and stop.