If we look at LMArena for example, GPT-5-Chat is ranked 20th, GPT-5.1 is 15th vs Opus 4.5 (non-thinking) is 3rd or Sonnet 10th. And just personally, I am very happy with GPT-5.1-Pro and Thinking, but I never use the instant version, to me it is quite sloppy and unreliable – not

→ View original post on X — @petergostev,