Resolves YES if any Claude 3 model outranks the best-performing GPT-4 model at any point within two weeks of first being listed, i.e. if users prefer Claude 3 responses to GPT-4 responses at any point.

(GPT-4.5 would not count as a GPT-4 model.)

https://huggingface.co/spaces/lmsys/chatbot-arena-leaderboard


EDIT: Clarification, due to the new way ranks are displayed on the LMSYS Chatbot Arena Elo Leaderboard: If Claude 3 and GPT-4 both have rank 1, but the Arena Elo of the former is greater than the one of the latter, this resolves YES. I believe this is most in line with how it was meant when this market was first created.