Frontier LLMs drop from 83% to 43% once reasoning has to chain across domains

Y	Hacker News new \| ask \| show \| jobs

	Frontier LLMs drop from 83% to 43% once reasoning has to chain across domains (arxiv.org)
	2 points by MarcoDewey 3 days ago

1 comments

Is that the model, all models, the agent?

Does something like this make a difference?

"Schema Harness Achieves ~99% on Arc‑AGI‑3 Public" (2026-07) https://news.ycombinator.com/item?id=48935905