LLMs 5
- Over the course of a few months, AI models have gotten much better at long-term coherence, multi-agent coordination and strategic reasoning.
- A team mixing two different frontier models performed, if anything, worse than teams of either model alone, though not significantly.