Research

Paper

AI LLM March 11, 2026

Ranking Reasoning LLMs under Test-Time Scaling

Authors

Mohsen Hariri, Michael Hinczewski, Jing Ma, Vipin Chaudhary

Abstract

Test-time scaling evaluates reasoning LLMs by sampling multiple outputs per prompt, but ranking models in this regime remains underexplored. We formalize dense benchmark ranking under test-time scaling and introduce Scorio, a library that implements statistical ranking methods such as paired-comparison models, item response theory (IRT) models, voting rules, and graph- and spectral-based methods. Across $20$ reasoning models on four Olympiad-style math benchmarks (AIME'24, AIME'25, HMMT'25, and BrUMO'25; up to $N=80$ trials), most full-trial rankings agree closely with the Bayesian gold standard $\mathrm{Bayes}_{\mathcal{U}}@80$ (mean Kendall's $τ_b = 0.93$--$0.95$), and $19$--$34$ methods recover exactly the same ordering. In the single-trial regime, the best methods reach $τ_b \approx 0.86$. Using greedy decoding as an empirical prior ($\mathrm{Bayes}_{\mathbf{R}_0}@N$) reduces variance at $N=1$ by $16$--$52\%$, but can bias rankings when greedy and stochastic sampling disagree. These results identify reliable ranking methods for both high- and low-budget test-time scaling. We release Scorio as an open-source library at https://github.com/mohsenhariri/scorio.

Metadata

arXiv ID: 2603.10960
Provider: ARXIV
Primary Category: cs.LG
Published: 2026-03-11
Fetched: 2026-03-12 04:21

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.10960v1</id>\n    <title>Ranking Reasoning LLMs under Test-Time Scaling</title>\n    <updated>2026-03-11T16:47:41Z</updated>\n    <link href='https://arxiv.org/abs/2603.10960v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.10960v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Test-time scaling evaluates reasoning LLMs by sampling multiple outputs per prompt, but ranking models in this regime remains underexplored. We formalize dense benchmark ranking under test-time scaling and introduce Scorio, a library that implements statistical ranking methods such as paired-comparison models, item response theory (IRT) models, voting rules, and graph- and spectral-based methods. Across $20$ reasoning models on four Olympiad-style math benchmarks (AIME'24, AIME'25, HMMT'25, and BrUMO'25; up to $N=80$ trials), most full-trial rankings agree closely with the Bayesian gold standard $\\mathrm{Bayes}_{\\mathcal{U}}@80$ (mean Kendall's $τ_b = 0.93$--$0.95$), and $19$--$34$ methods recover exactly the same ordering. In the single-trial regime, the best methods reach $τ_b \\approx 0.86$. Using greedy decoding as an empirical prior ($\\mathrm{Bayes}_{\\mathbf{R}_0}@N$) reduces variance at $N=1$ by $16$--$52\\%$, but can bias rankings when greedy and stochastic sampling disagree. These results identify reliable ranking methods for both high- and low-budget test-time scaling. We release Scorio as an open-source library at https://github.com/mohsenhariri/scorio.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.LG'/>\n    <category scheme='http://arxiv.org/schemas/atom' term='math.ST'/>\n    <published>2026-03-11T16:47:41Z</published>\n    <arxiv:comment>Code is available at https://github.com/mohsenhariri/scorio</arxiv:comment>\n    <arxiv:primary_category term='cs.LG'/>\n    <author>\n      <name>Mohsen Hariri</name>\n    </author>\n    <author>\n      <name>Michael Hinczewski</name>\n    </author>\n    <author>\n      <name>Jing Ma</name>\n    </author>\n    <author>\n      <name>Vipin Chaudhary</name>\n    </author>\n  </entry>"
}