OpenDeepThink: Parallel Reasoning via Bradley-Terry Aggregation

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Zhou, Shang, Chai, Wenhao, Liu, Kaiyuan, Mao, Huanzhi, Mang, Qiuyang, Shang, Jingbo
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866916019973390336
author Zhou, Shang
Chai, Wenhao
Liu, Kaiyuan
Mao, Huanzhi
Mang, Qiuyang
Shang, Jingbo
author_facet Zhou, Shang
Chai, Wenhao
Liu, Kaiyuan
Mao, Huanzhi
Mang, Qiuyang
Shang, Jingbo
contents Test-time compute scaling is a primary axis for improving LLM reasoning. Existing methods primarily scale depth by extending a single reasoning trace. Scaling breadth by sampling multiple candidates in parallel is straightforward, but introduces a selection bottleneck: choosing the best candidate without a ground-truth verifier, since pointwise LLM judging is noisy and biased. To address this, we introduce OpenDeepThink, a population-based test-time compute framework that selects via pairwise Bradley-Terry comparison. Each generation, the LLM judges random pairs of candidates and aggregates votes via Bradley-Terry into a global ranking; top-ranked candidates are preserved and the top three quarters are mutated using the natural-language critiques produced during comparison; the bottom quarter is discarded. OpenDeepThink raises Gemini 3.1 Pro's effective Codeforces Elo by +405 points in eight sequential LLM-call rounds (~27 minutes wall-clock). The pipeline transfers across weaker and stronger models without retuning, and on the multi-domain HLE benchmark, gains appear concentrated in objectively verifiable domains and reverse in subjective ones. We release CF-73, a curated set of 73 expert-rated Codeforces problems with International Grandmaster annotation and 99% local-evaluation agreement against the official verdict.
format Preprint
id arxiv_https___arxiv_org_abs_2605_15177
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle OpenDeepThink: Parallel Reasoning via Bradley-Terry Aggregation
Zhou, Shang
Chai, Wenhao
Liu, Kaiyuan
Mao, Huanzhi
Mang, Qiuyang
Shang, Jingbo
Artificial Intelligence
Test-time compute scaling is a primary axis for improving LLM reasoning. Existing methods primarily scale depth by extending a single reasoning trace. Scaling breadth by sampling multiple candidates in parallel is straightforward, but introduces a selection bottleneck: choosing the best candidate without a ground-truth verifier, since pointwise LLM judging is noisy and biased. To address this, we introduce OpenDeepThink, a population-based test-time compute framework that selects via pairwise Bradley-Terry comparison. Each generation, the LLM judges random pairs of candidates and aggregates votes via Bradley-Terry into a global ranking; top-ranked candidates are preserved and the top three quarters are mutated using the natural-language critiques produced during comparison; the bottom quarter is discarded. OpenDeepThink raises Gemini 3.1 Pro's effective Codeforces Elo by +405 points in eight sequential LLM-call rounds (~27 minutes wall-clock). The pipeline transfers across weaker and stronger models without retuning, and on the multi-domain HLE benchmark, gains appear concentrated in objectively verifiable domains and reverse in subjective ones. We release CF-73, a curated set of 73 expert-rated Codeforces problems with International Grandmaster annotation and 99% local-evaluation agreement against the official verdict.
title OpenDeepThink: Parallel Reasoning via Bradley-Terry Aggregation
topic Artificial Intelligence
url https://arxiv.org/abs/2605.15177