logoalt Hacker News

apitmanyesterday at 5:36 PM1 replyview on HN

Are you sure the difference is from MoE and not that 3.6 is newer?


Replies

riknos314yesterday at 6:50 PM

Qwen 3.5 27B also scores higher than 3.5 122BA10B. So even in the same generation the smaller dense model outperformed the larger MOE

show 1 reply