Commentary on X
same energy there's no hard qualitative boundary between "Erdos problem" and "Putnam problem" OpenAI just invests into doing these publicity stunts I would obviously love it if DeepSeek claimed some *first success* on a named problem, and the frontier is ahead but it's nbd
8:03 AM · Jul 15, 2026> Mind you, DeepSeek-V4, GLM-5.2 or Kimi-K2.6/2.7 haven't solved a single Erdös problem so far. Ok this is enough Lisan. Mind you, not all Erdös problems are equally hard Keep trying tho!
7:49 AM · Jul 15, 2026@scaling01 > Mind you, DeepSeek-V4, GLM-5.2 or Kimi-K2.6/2.7 haven't solved a single Erdös problem so far. this is probably not true but i agree that there is a lack of evidence here, is there any public benchmark that track this?
7:44 AM · Jul 15, 2026View all
@eliebakouch not really, I personally haven't heard anything of this sort and im chronically online and I send out a GPT-5.6-xhigh request to check for exactly this
7:48 AM · Jul 15, 2026