Is GPT-4 getting dumber? We may finally have some proof

Researchers claim that OpenAI's GPT-4 language model is not only performing worst than its prior GPT 3.5 iteration but from older GPT-4 versions.