The authors find collaboration is non-monotonic: peer models can rescue failures no single model solves, but can also corrupt answers that were initially correct.
COMED combines anchor self-consistency, router margin and a lightweight peer probe to accept confident answers, verify ambiguous ones, and escalate only when collaboration looks beneficial.
Across 16 open weightsA model whose trained parameters are published for anyone to download and run — unlike API-only models you can access but never possess.Full definition → settings it improved both fixed and routed anchors, gaining up to 10.7 points on MedQA while invoking fewer models and decoding fewer tokenThe chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.Full definition → than dense collaboration.
With frontier models on HLE, COMED lifted GPT-5.5 from 23.1% to 28.1%, beating dense collaboration.
Terms in this piece · Glossary
inference — Running a trained model to get answers — the phase where AI is actually used, as opposed to trained.
open weights — A model whose trained parameters are published for anyone to download and run — unlike API-only models you can access but never possess.
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
Why it matters
Shows how to decide when to invoke peer models instead of always routing or always collaborating, directly useful for anyone building multi-model inferenceRunning a trained model to get answers — the phase where AI is actually used, as opposed to trained.Full definition → pipelines.
articleWhy Better Models Can Create Riskier Systems: Evidence from LLM Agents in Financial MarketsJillian Ross, Eric So, Zoe De Simone, Charles Pozniak, Andrew W. Lo
articleCaught in the Story: Narrative Captivity in Multi-turn LLMs ConversationYuhe Wu, Guangyu Wang, Yujie Chen, Jiatong Zhang, Yuran Chen, Yutong Zhang, Xiyin Cheng, Wenpeng Cao, Zhuang Liu, Guang Zhang
articleSciReC: Diagnostic Evaluation of Multimodal, Multi-Turn Relational Reasoning with Adaptive InteractionNilay Yilmaz, Naga Sai Abhiram Kusumba, Stella Wenxing Liu, Yezhou Yang