Do We Need Frontier Models to Verify Mathematical Proofs?

arXiv cs.AI / 4/6/2026

💬 OpinionSignals & Early TrendsIdeas & Deep AnalysisModels & Research

Key Points

  • The paper evaluates how well frontier and open-source LLMs can verify natural-language mathematical proofs, using verifier accuracy and self-consistency as key metrics.
  • Results show smaller open-source models are close to frontier models in accuracy (within ~10%) but are markedly less consistent across repeated judgments (up to ~25% worse).
  • Verifier accuracy is highly sensitive to prompt choice for all models, indicating that “verification” reliability depends not only on model capability but also on elicitation strategy.
  • The authors find that smaller models can verify at frontier-level capability, but general judging prompts fail to reliably trigger those abilities.
  • An LLM-guided prompt search produces an ensemble of specialized prompts that improves smaller models’ accuracy by up to 9.1% and self-consistency by up to 15.9%, enabling models like Qwen3.5-35B to match frontier models (e.g., Gemini 3.1 Pro) on proof verification.

Abstract

Advances in training, post-training, and inference-time methods have enabled frontier reasoning models to win gold medals in math competitions and settle challenging open problems. Gaining trust in the responses of these models requires that natural language proofs be checked for errors. LLM judges are increasingly being adopted to meet the growing demand for evaluating such proofs. While verification is considered easier than generation, what model capability does reliable verification actually require? We systematically evaluate four open-source and two frontier LLMs on datasets of human-graded natural language proofs of competition-level problems. We consider two key metrics: verifier accuracy and self-consistency (the rate of agreement across repeated judgments on the same proof). We observe that smaller open-source models are only up to ~10% behind frontier models in accuracy but they are up to ~25% more inconsistent. Furthermore, we see that verifier accuracy is sensitive to prompt choice across all models. We then demonstrate that the smaller models, in fact, do possess the mathematical capabilities to verify proofs at the level of frontier models, but they struggle to reliably elicit these capabilities with general judging prompts. Through an LLM-guided prompt search, we synthesize an ensemble of specialized prompts that overcome the specific failure modes of smaller models, boosting their performance by up to 9.1% in accuracy and 15.9% in self-consistency. These gains are realized across models and datasets, allowing models like Qwen3.5-35B to perform on par with frontier models such as Gemini 3.1 Pro for proof verification.