#sequential decision making

LLM Agents Face Scale Problems on Sequential Decision Tasks

New benchmarks and multi-agent systems expose performance gaps when language models must reason through long chains of decisions.