>>106156027
Anyone who believed this benchmeme just because they like fancy charts should neck themselves, in the first place. Believing LLM judged tests and n-gram analysis is delusional. The dev behind it is an absolute moron who also created the "slop test" that shows GPT-4.5 and Grok 3 Beta to be sibling models (lmao). He also repeatedly made inane claims, like that model X was trained on model Y while clearly having never interacted with either.