When Combining Language Models Stops Helping

This episode explores Josef Chen’s paper on when combining language models actually improves accuracy, focusing on the difference between pairwise error correlation and the more decisive co-failure rate, beta: the chance that every model in a pool fails on the same query. It explains why beta sets a hard ceiling for routing, voting, cascades, and post-training Mixture-of-Agents systems, and why the real gain over a strong single model only exists on queries where that model fails but another succeeds. The discussion walks through results from a 15-model routing setup and a 67-model frontier-model study, showing that even calibrated copula-based estimates systematically understate shared failure and that learned routers capture only a small fraction of the available oracle gain. A listener would find it interesting because it cuts through ensemble hype with a concrete argument about when multi-model orchestration is worth the added cost and complexity, plus a practical way to estimate headroom before building a router at all. Sources: 1. When Does Combining Language Models Help? A Co-Failure Ceiling on Routing, Voting, and Mixture-of-Agents Across 67 Frontier Models — Josef Chen, 2026 http://arxiv.org/abs/2606.27288 2. LLM-Blender: Ensembling Large Language Models with Pairwise Ranking and Generative Fusion — Dongfu Jiang, Xiang Ren, Bill Yuchen Lin, 2023 https://scholar.google.com/scholar?q=LLM-Blender%3A+Ensembling+Large+Language+Models+with+Pairwise+Ranking+and+Generative+Fusion 3. Mixture-of-Agents Enhances Large Language Model Capabilities — Junlin Wang, Jue Wang, Ben Athiwaratkun, Ce Zhang, James Zou, 2024 https://scholar.google.com/scholar?q=Mixture-of-Agents+Enhances+Large+Language+Model+Capabilities 4. Rethinking Mixture-of-Agents: Is Mixing Different Large Language Models Beneficial? — Wenzhe Li, Yong Lin, Mengzhou Xia, Chi Jin, 2025 https://scholar.google.com/scholar?q=Rethinking+Mixture-of-Agents%3A+Is+Mixing+Different+Large+Language+Models+Beneficial%3F 5. When Does Combining Language Models Help? A Co-Failure Ceiling on Routing, Voting, and Mixture-of-Agents Across 67 Frontier Models — Josef Chen, 2026 https://scholar.google.com/scholar?q=When+Does+Combining+Language+Models+Help%3F+A+Co-Failure+Ceiling+on+Routing%2C+Voting%2C+and+Mixture-of-Agents+Across+67+Frontier+Models 6. A Unified Approach to Routing and Cascading for LLMs — Jasper Dekoninck, Maximilian Baader, and Martin Vechev, 2024 https://scholar.google.com/scholar?q=A+Unified+Approach+to+Routing+and+Cascading+for+LLMs 7. When Does Confidence-Based Cascade Deferral Suffice? — Wittawat Jitkrittum, Neha Gupta, Aditya Krishna Menon, Harikrishna Narasimhan, Ankit Singh Rawat, and Sanjiv Kumar, 2023 https://scholar.google.com/scholar?q=When+Does+Confidence-Based+Cascade+Deferral+Suffice%3F 8. Correlated Errors in Large Language Models — Elliot Kim, Avi Garg, Kenny Peng, and Nikhil Garg, 2025 https://scholar.google.com/scholar?q=Correlated+Errors+in+Large+Language+Models 9. Don't Always Pick the Highest-Performing Model: An Information-Theoretic View of LLM Ensemble Selection — Yigit Turkmen, Baturalp Buyukates, and Melih Bastopcu, 2026 https://scholar.google.com/scholar?q=Don%27t+Always+Pick+the+Highest-Performing+Model%3A+An+Information-Theoretic+View+of+LLM+Ensemble+Selection 10. PAG: Multi-Turn Reinforced LLM Self-Correction with Policy as Generative Verifier — Yuhua Jiang et al., 2025 https://arxiv.org/abs/2506.10406 11. S2R: Teaching LLMs to Self-verify and Self-correct via Reinforcement Learning — Ruotian Ma et al., 2025 https://arxiv.org/abs/2502.12853 12. Small Language Models Need Strong Verifiers to Self-Correct Reasoning — Yunxiang Zhang et al., 2024 https://arxiv.org/abs/2404.17140 13. CP-Router: An Uncertainty-Aware Router Between LLM and LRM — Jiayuan Su et al., 2025 https://arxiv.org/abs/2505.19970 14. Leveraging Uncertainty Estimation for Efficient LLM Routing — Tuo Zhang et al., 2025 https://arxiv.org/abs/2502.11021 15. Confident or Seek Stronger: Exploring Uncertainty-Based On-device LLM Routing From Benchmarking to Generalization — Yu-Neng Chuang et al., 2025 https://arxiv.org/abs/2502.04428 16. Wisdom and Delusion of LLM Ensembles for Code Generation and Repair — Fernando Vallecillos Ruiz et al., 2025 https://arxiv.org/abs/2510.21513 17. Understanding Agent Scaling in LLM-Based Multi-Agent Systems via Diversity — Yingxuan Yang et al., 2026 https://arxiv.org/abs/2602.03794 18. LLM Chemistry Estimation for Multi-LLM Recommendation — Huascar Sanchez and Briland Hitaj, 2025 https://arxiv.org/abs/2510.03930 19. AI Post Transformers: TMAS: Scaling Test-Time Compute with Multi-Agent Synergy — Hal Turing & Dr. Ada Shannon, 2026 https://podcast.do-not-panic.com/episodes/2026-05-14-tmas-scaling-test-time-compute-with-mult-3abe7a.mp3 20. AI Post Transformers: IMO-Bench for Robust Mathematical Reasoning — Hal Turing & Dr. Ada Shannon, 2026 https://podcast.do-not-panic.com/episodes/2026-04-04-imo-bench-for-robust-mathematical-reason-143489.mp3