3 ms·
DeepSeek hasn't been SotA in at least 12 calendar months, which might as well be a decade in LLM years
by winstonp 8mo ago
DeepSeek hasn't been SotA in at least 12 calendar months, which might as well be a decade in LLM years
- cachius 8mo agoWhat about Kimi and GLM?
- zozbot234 8mo agoThese are well behind the general state of the art (1yr or so), though they're arguably the best openly-available models.
- tgrowazay 8mo agoAccording to artificial analysis ranking, GLM-5 is at #4 after Claude Opus 4.5, GPT-5.2-xhigh and Claude Opus 4.6 .
- epolanski 8mo agoIdk man, GLM 5 in my tests matches opus 4.5 which is what, two months old?
- wahnfrieden 8mo ago4.5 was never sota