Commentary on X
@scaling01 dude, i *love* this model. no idea if it can pull it off but it has a lovely way about it.
9:34 AM · Jul 16, 2026@scaling01 whenever they do that I just hate them, if they need encouragement means somethings not right, a helper can't be a useless shy little bitch
11:01 AM · Jul 16, 2026I would actually love Moonshot (or any other open lab) using such a proof as their headline result for the launch if they can solve any relevant problem
9:42 AM · Jul 16, 2026@scaling01 a model failing on a hard problem and someone actually reading the cot is the whole point of doing this in public. good work.
9:42 AM · Jul 16, 2026View all
@scaling01 all I see here is rlhf ptsd
11:04 AM · Jul 16, 2026Kimi-K3 didn't solve the unit distance problem GPT-5.6-Sol's assessment of Kimi's raw CoTs
9:40 AM · Jul 16, 2026lots of post-training it will not even attempt to solve a hard problem
9:00 AM · Jul 16, 2026