The authors constructed RouterWorld, a comprehensive benchmark for LLM routing in conversational AI.
New benchmark shows LLM routers ignore latency when feedback is incomplete
RouterWorld evaluates 13 routing strategies under partial observability, finding most trade efficiency for accuracy.
Big Tech
Jinyang Li · Denis Emelin · Simone Tedeschi · Weiwei Cheng · Cristina Morariu · Haytham Assem · +3 more
The University of Hong Kong · Amazon · University College London
Research Digest··3 min read
The authors introduce RouterWorld, a benchmark that reunifies 371k single-turn responses and 76k multi-turn agentic trajectories across seven open-weight LLMs, and formalizes partial observability via Cold-Start and Voyager setups.
§