Re:Qwen4-Max
by zkmon on Monday November 16, 2026 @01:47AM
(#50514759)
Why do we trust this? Do they open-source their test code?
Also, a generic benchmark might not translate well into what everyone needs. I keep building my own test cases that test models for my own work specifics. Only that benchmark is useful to me. I don't need it to win a PhD math quiz.
Also, a generic benchmark might not translate well into what everyone needs. I keep building my own test cases that test models for my own work specifics. Only that benchmark is useful to me. I don't need it to win a PhD math quiz.