Metraipolis › archive › dc_407b68cac150

COMED: The Missing Middle Between Routing and Collaboration in Multi-LLM Inference — arXiv:2609.26913

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.26913
License (as archived)
CC0-1.0
Archived
2026-10-08T09:11:59.950Z
SHA-256 of the archived text
87b8739e39cab702ad3435e50ae8ff248b8a4a3cb6b998901c05c800812bb170
Stable URL of this page
https://metraipolis.com/documento/dc_407b68cac150

How to cite

COMED: The Missing Middle Between Routing and Collaboration in Multi-LLM Inference — arXiv:2609.26913 — arXiv (en), https://arxiv.org/abs/2609.26913 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_407b68cac150 (Metraipolis archive dc_407b68cac150, sha256 87b8739e39cab702ad3435e50ae8ff248b8a4a3cb6b998901c05c800812bb170)
BibTeX
@misc{dc407b68cac150,
  title = {COMED: The Missing Middle Between Routing and Collaboration in Multi-LLM Inference — arXiv:2609.26913},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.26913},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_407b68cac150; sha256 87b8739e39cab702ad3435e50ae8ff248b8a4a3cb6b998901c05c800812bb170},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.26913. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

No single Large Language Model (LLM) is uniformly reliable across queries, motivating multi-model inference systems that either route among models or combine their outputs. However, routing stops after selecting an initial model, while dense collaboration invokes peers on every query. We show that collaboration is non-monotonic: peers can recover failures that no model solves alone, but can also corrupt initially correct answers. We introduce COMED (Controlled Model Escalation for Multi-LLM Deliberation), a post-anchor controller for selective cross-model collaboration. COMED uses anchor self-consistency, router margin, and a lightweight peer probe to accept confident answers, verify ambiguous cases, and escalate only when collaboration is likely beneficial. We formalize this trade-off with a rescue-harm decomposition showing that selective collaboration improves when rescued errors outweigh collaboration-induced harms. Across medical, scientific, and general reasoning benchmarks, COMED improves fixed and routed anchors in all 16 open-weight settings, with gains up to +10.7 percentage points on MedQA while invoking fewer models and using fewer decoded tokens than dense collaboration. On HLE with frontier models, COMED improves GPT-5.5 from 23.1% to 28.1%, outperforming dense collaboration and achieving the best results.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-24 in cs.AI).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.26913 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-08T08:13:26.594Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-24?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).