mlsanityData to 14 Oct 2026

Best-of-∞ -- Asymptotic Performance of Test-Time LLM Ensembling

arXiv 2509.21091 ↗First posted 25 Sep 2025stat.ML cs.AI cs.LGcited by 3

Not on the map. Why?

Abstract (arXiv)

We study best-of-N for large language models (LLMs) where the selection is based on majority voting. In particular, we analyze the limit N → ∞, which we denote as boinflower. While this approach achieves impressive performance in the limit, it requires an infinite test-time budget. To address this, we propose an adaptive generation scheme that selects N based on answer agreement, thereby efficiently allocating inference-time computation. Beyond adaptivity, we extend the framework to weighted ensembles of multiple LLMs, showing that such mixtures can outperform any individual model. The optimal ensemble weighting is formulated and efficiently computed as a mixed-integer linear program. Extensive experiments demonstrate the effectiveness of our approach.

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 50 of its 77 references that are in the map. First posted on arXiv 25 Sep 2025.

Builds on

50 of 77 references.

References · 50

  1. 2023GPT-4 Technical Reportcited by 18,396
  2. 2024The Llama 3 Herd of Modelscited by 14,715
  3. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  4. 2021Training Verifiers to Solve Math Word Problemscited by 10,109
  5. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  6. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,708
  7. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,371
  8. 2022Large Language Models are Zero-Shot Reasonerscited by 4,374
  9. 2025Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilitiescited by 3,796
  10. 2023GPQA: A Graduate-Level Google-Proof Q&A Benchmarkcited by 2,879
  11. 2022BLOOM: A 176B-Parameter Open-Access Multilingual Language Modelcited by 1,984
  12. 2023Pythia: A Suite for Analyzing Large Language Models Across Training and Scalingcited by 1,786
38 more
  1. 2025s1: Simple test-time scalingcited by 1,341
  2. 2024Large Language Monkeys: Scaling Inference Compute with Repeated Samplingcited by 991
  3. 2022Solving math word problems with process- and outcome-based feedbackcited by 901
  4. 2023The Falcon Series of Open Language Modelscited by 535
  5. 2025LIMO: Less is More for Reasoningcited by 498
  6. 2024Skywork-Reward: Bag of Tricks for Reward Modeling in LLMscited by 327
  7. 2023Alphazero-like Tree-Search can Guide Large Language Model Decoding and Trainingcited by 295
  8. 2024Inference Scaling Laws: An Empirical Analysis of Compute-Optimal Inference for Problem-Solving with Language Modelscited by 290
  9. 2024RLHF Workflow: From Reward Modeling to Online RLHFcited by 267
  10. 2016Optimal Best Arm Identification with Fixed Confidencecited by 239
  11. 2025Learning to Reason without External Rewardscited by 210
  12. 2025Deep Think with Confidencecited by 205
  13. 2024More Agents Is All You Needcited by 180
  14. 2024Secrets of RLHF in Large Language Models Part II: Reward Modelingcited by 171
  15. 2025RM-R1: Reward Modeling as Reasoningcited by 154
  16. 2025Phi-4-reasoning Technical Reportcited by 151
  17. 2024Generative Reward Modelscited by 148
  18. 2023Large Language Model Cascades with Mixture of Thoughts Representations for Cost-efficient Reasoningcited by 143
  19. 2023Let's Sample Step by Step: Adaptive-Consistency for Efficient Reasoning and Coding with LLMscited by 115
  20. 2025NVIDIA Nemotron Nano 2: An Accurate and Efficient Hybrid Mamba-Transformer Reasoning Modelcited by 108
  21. 2023LLM360: Towards Fully Transparent Open-Source LLMscited by 104
  22. 2025Is Best-of-N the Best of Them? Coverage, Scaling, and Optimality in Inference-Time Alignmentcited by 97
  23. 2015Parallelizing the dual revised simplex methodcited by 89
  24. 2025Harnessing Multiple Large Language Models: A Survey on LLM Ensemblecited by 86
  25. 2018Mixture Martingales Revisited with Applications to Sequential Tests and Confidence Intervalscited by 76
  26. 2025InftyThink: Breaking the Length Limits of Long-Context Reasoning in Large Language Modelscited by 54
  27. 2025Reward Reasoning Modelcited by 47
  28. 2025The Majority is not always right: RL training for solution aggregationcited by 47
  29. 2022Model Cascading: Towards Jointly Improving Efficiency and Accuracy of NLP Systemscited by 44
  30. 2023Getting MoRE out of Mixture of Language Model Reasoning Expertscited by 44
  31. 2024LLM-TOPLA: Efficient LLM Ensemble by Maximising Diversitycited by 42
  32. 2024Smoothie: Label Free Language Model Routingcited by 31
  33. 2025AM-Thinking-v1: Advancing the Frontier of Reasoning at 32B Scalecited by 29
  34. 2025EXAONE Deep: Reasoning Enhanced Language Modelscited by 21
  35. 2025K2-Think: A Parameter-Efficient Reasoning Systemcited by 11
  36. 2025DynScaling: Efficient Verifier-free Inference Scaling via Dynamic and Integrated Samplingcited by 7
  37. 2020Anytime-Valid Inference for Multinomial Count Datacited by 5
  38. 2025Test-Time Scaling with Reflective Generative Modelcited by 4

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 1; 2026: 2.

3 papers.

  1. 2025Re-FORC: Adaptive Reward Prediction for Efficient Chain-of-Thought Reasoningcited by 4
  2. 2026Test-Time Compute Gamescited by 4
  3. 2026Strategic Self-Consistencycited by 0