← All news

Research

Preprint proposes pairwise signals to tighten LLM math rankings

A revised preprint reports a method intended to make mathematical-reasoning evaluations of large language models more precise when benchmarks are small or model outputs vary between runs.

reviewedUpdated Sep 4, 2026, 9:02 PM UTC
Original source

arXiv

Read the original source

What happened

The July 20, 2026 revision of “Evaluating LLMs When They Do Not Know the Answer” proposes combining ordinary accuracy results with model judgments that compare two auxiliary reasoning chains.

The authors treat those comparisons as auxiliary signals in a statistical estimator. Rather than relying only on whether a final answer is correct, the method uses judgments about which of two candidate solutions is better.

In author-run simulations and tests on GPQA Diamond, AIME 2025, and GSM8K, the paper reports more precise performance estimates and generally more reliable model rankings than naive sample averaging. The reported gains were strongest with small evaluation samples.

AIME 2025 has 30 problems, so changing one item moves raw accuracy by more than three percentage points. That makes small score differences especially coarse on that benchmark.

Why it matters

Benchmark rankings often guide claims about which models are stronger. Separate research has found uncertainty in benchmark scores and substantial ranking variability, but it did not test this paper’s method. More precise estimates of a fixed score also do not show that a benchmark measures reasoning validity.

What remains unclear

Sources

  1. Evaluating LLMs When They Do Not Know the Answer: Statistical Evaluation of Mathematical Reasoning via Comparative SignalsPrimary source - arXiv - preprint paper; subject-authored research - Jul 20, 2026

    Used for: The revised preprint, its proposed estimator, benchmark scope, and reported results.

    Open source

  2. Towards Reproducible LLM Evaluation: Quantifying Uncertainty in LLM Benchmark ScoresThe Alan Turing Institute - institution-hosted preprint; independent academic research - Date not listed

    Used for: Context on uncertainty in LLM benchmark scores.

    Open source

  3. Quantifying Ranking Uncertainty in LLM BenchmarksarXiv - preprint paper; independent academic research - Jun 28, 2026

    Used for: Context on ranking variability in LLM benchmarks.

    Open source