Experimental 3.5T K2 merge beats GPT-4.5 and Opus at writinghuggingface.co 2 pointsbo0tzz10 months ago1 commentSaveHideCopy link On HNComments−mynti10mo"beats at writing" is kind of far fetched since the page provides three generic pormpts with answers from Opus, GPT5 and K3. But a 3.5T model is crazy huge and the answers still feel more or less the same to the other models
Comments
"beats at writing" is kind of far fetched since the page provides three generic pormpts with answers from Opus, GPT5 and K3. But a 3.5T model is crazy huge and the answers still feel more or less the same to the other models