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

2 points by MarcoDewey 26 days ago | 1 comment

westurner 26 days ago |

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

westurner 26 days ago |

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